Merge pull request #260 from CSCfi/CSCMETAX-398-allow-file-uptade-from-0-to-n

CSCMETAX-398: [ADD] Datasets api: Allow the first addition of files t…

Author: junsk1

src/metax_api/models/catalog_record.py

@@ -788,7 +788,14 @@ def _pre_update_operations(self):
                         'back to 0.' % self.preservation_state
                     ]})
 
-                self._create_new_dataset_version()
+                if self._files_added_for_first_time():
+                    # first update from 0 to n files should not create a dataset version. all later updates
+                    # will create new dataset versions normally.
+                    self.files.add(*self._get_dataset_selected_file_ids())
+                    self._calculate_total_ida_byte_size()
+                    self._handle_metadata_versioning()
+                else:
+                    self._create_new_dataset_version()
 
             else:
                 if self.preservation_state in (
@@ -822,6 +829,29 @@ def _pre_update_operations(self):
             if self.catalog_is_harvested() and self.field_changed('research_dataset.preferred_identifier'):
                 self._handle_preferred_identifier_changed()
 
+    def _files_added_for_first_time(self):
+        """
+        Find out if this update is the first time files are being added/changed since the dataset's creation.
+        """
+        if self.files.exists():
+            # current version already has files
+            return False
+
+        if self.dataset_version_set.records.count() > 1:
+            # for versioned catalogs, when a record is first created, the record is appended
+            # to dataset_version_set. more than one dataset versions existing implies files
+            # have changed already in the past.
+            return False
+
+        metadata_versions_with_files_exist = ResearchDatasetVersion.objects.filter(
+            Q(Q(research_dataset__files__isnull=False) | Q(research_dataset__directories__isnull=False)),
+            catalog_record_id=self.id) \
+            .exists()
+
+        # metadata_versions_with_files_exist == True implies this "0 to n" update without
+        # creating a new dataset version already occurred once
+        return not metadata_versions_with_files_exist
+
     def _calculate_total_ida_byte_size(self):
         rd = self.research_dataset
         if 'files' in rd or 'directories' in rd:
@@ -964,6 +994,7 @@ def _files_changed(self):
                 temp_record.id = None
                 temp_record.next_dataset_version = None
                 temp_record.previous_dataset_version = None
+                temp_record.dataset_version_set = None
                 temp_record.identifier = generate_identifier()
                 temp_record.research_dataset['metadata_version_identifier'] = generate_identifier()
                 super(Common, temp_record).save()
@@ -1035,7 +1066,7 @@ def _create_new_dataset_version(self):
         new_version.service_created = old_version.service_modified or old_version.service_created
         new_version.service_modified = None
         new_version.alternate_record_set = None
-        new_version.dataset_version_set.records.add(new_version)
+        old_version.dataset_version_set.records.add(new_version)
 
         # note: copying research_dataset from the currently open instance 'old_version',
         # contains the new field data from the request. this effectively transfers

src/metax_api/tests/api/rest/base/views/datasets/write.py

@@ -1376,6 +1376,48 @@ class CatalogRecordApiWriteDatasetVersioning(CatalogRecordApiWriteCommon):
     Catalogs 1-2 should have dataset_versioning=True, while the rest should not.
     """
 
+    def test_update_from_0_to_n_files_does_not_create_new_version(self):
+        """
+        The FIRST update from 0 to n files in a dataset should be permitted
+        without creating a new dataset version.
+        """
+        data = self.client.get('/rest/datasets/1', format="json").data
+        data.pop('id')
+        data.pop('identifier')
+        data['research_dataset'].pop('preferred_identifier', None)
+        files = data['research_dataset'].pop('files', None)
+        data['research_dataset'].pop('directories', None)
+        self.assertEqual(isinstance(files, list), True)
+
+        # create test record
+        response = self.client.post('/rest/datasets', data, format="json")
+        self.assertEqual(response.status_code, status.HTTP_201_CREATED, response.data)
+
+        # modify a few times to create metadata versions
+        data = response.data
+        data['research_dataset']['title']['en'] = 'updated'
+        data = self.client.put('/rest/datasets/%d' % data['id'], data, format="json").data
+        data['research_dataset']['title']['en'] = 'updated again'
+        data = self.client.put('/rest/datasets/%d' % data['id'], data, format="json").data
+
+        # add files for the first time - should not create a new dataset version
+        data['research_dataset']['files'] = files
+        response = self.client.put('/rest/datasets/%d' % data['id'], data, format="json")
+        self.assertEqual(response.status_code, status.HTTP_200_OK, response.data)
+        self.assertEqual('new_version_created' in response.data, False)
+
+        # remove files again... a new version is created normally
+        files = data['research_dataset'].pop('files')
+        response = self.client.put('/rest/datasets/%d' % data['id'], data, format="json")
+        new_version = self.get_next_version(response.data)
+
+        # ...and put the files back. this is another 0->n files update. this time
+        # should normally create new dataset version.
+        new_version['research_dataset']['files'] = files
+        response = self.client.put('/rest/datasets/%d' % new_version['id'], new_version, format="json")
+        self.assertEqual(response.status_code, status.HTTP_200_OK, response.data)
+        self.assertEqual('new_version_created' in response.data, True)
+
     def test_update_to_non_versioning_catalog_does_not_create_version(self):
         self._set_cr_to_catalog(pk=self.pk, dc=3)
         response = self._get_and_update_title(self.pk)