VERSION

-0.10.4-rc1
+0.10.4-rc2

arkindex/dataimport/models.py

@@ -122,6 +122,11 @@ class DataImport(IndexableModel):
             if elements:
                 command = ' '.join([command, '--dataimport-id {}'.format(str(self.id))])
 
+            # Do not retrieve elements children in case there of thumbnails generation only
+            thumbnails = self.payload.get('thumbnails')
+            if thumbnails and len(self.ml_tools) == 0:
+                command = ' '.join([command, '--skip-children'.format(str(self.id))])
+
             tasks = {
                 import_task_name: {
                     'image': settings.ARKINDEX_TASKS_IMAGE,

arkindex/dataimport/tests/test_workflows_api.py

@@ -499,3 +499,31 @@ class TestWorkflows(FixtureAPITestCase):
             'python -m arkindex_tasks.init_elements --corpus-id {} --chunks-number 1 --dataimport-id {}'
             .format(str(self.corpus.id), str(dataimport.id))
         )
+
+    def test_thumbnails_generation_only(self, ml_get_mock):
+        """
+        Generating thumbnails without any workflow must generate an import task
+        which do not to retrieve filtered elements children
+        """
+        self.client.force_login(self.user)
+        corpus_type = self.corpus.types.first()
+        response = self.client.post(
+            reverse('api:corpus-workflow'),
+            {
+                'ml_tools': [],
+                'corpus': str(self.corpus.id),
+                'chunks': 3,
+                'type': corpus_type.slug,
+                'thumbnails': True
+            },
+            format='json'
+        )
+        self.assertEqual(response.status_code, status.HTTP_201_CREATED)
+        data = response.json()
+        dataimport = DataImport.objects.get(id=data['id'])
+        workflow = dataimport.workflow
+        self.assertEqual(
+            workflow.recipes['initialisation'].command,
+            'python -m arkindex_tasks.init_elements --corpus-id {} --chunks-number 3 --type {} --skip-children'
+            .format(str(self.corpus.id), corpus_type.slug)
+        )

arkindex/images/management/commands/reorder_polygons.py

@@ -38,6 +38,7 @@ class Command(BaseCommand):
             '--update-batch-size',
             help='Batch size used to update zones. Updates are run with each retrieval batch.',
             type=int,
+            default=1000,
         )
         parser.add_argument(
             '--dry-run',
@@ -51,7 +52,15 @@ class Command(BaseCommand):
             default=True,
         )
 
-    def handle(self, *args, offset=0, limit=0, batch_size=1000, update_batch_size=None, remove=True, **kwargs):
+    def handle(self,
+               *args,
+               offset=0,
+               limit=0,
+               batch_size=1000,
+               update_batch_size=1000,
+               dry_run=False,
+               remove=True,
+               **kwargs):
         assert batch_size > 0, 'Batch size must be positive.'
         assert update_batch_size > 0, 'Update batch size must be positive.'
 
@@ -77,10 +86,14 @@ class Command(BaseCommand):
                     zone.polygon.reorder()
                 except Exception as e:
                     if old_polygon.width < 1 or old_polygon.height < 1 and remove:
-                        logger.info('Removing incoherent zone {}…'.format(zone.id))
-                        zone.transcriptions.all().delete()
-                        zone.elements.all().delete()
-                        zone.delete()
+                        if dry_run:
+                            logger.info('Would remove incoherent zone {}'.format(zone.id))
+                        else:
+                            logger.info('Removing incoherent zone {}…'.format(zone.id))
+                            zone.transcriptions.all().delete()
+                            zone.elements.all().delete()
+                            zone.regions.all().delete()
+                            zone.delete()
                         removed += 1
                         total -= 1
                     else:
@@ -90,9 +103,29 @@ class Command(BaseCommand):
                     continue
 
                 if zone.polygon != old_polygon:
-                    reordered_zones.append(zone)
+                    # Check if a zone already exists and either reorder or deduplicate
+                    existing_zone = Zone.objects \
+                        .filter(image_id=zone.image_id, polygon=zone.polygon) \
+                        .exclude(id=zone.id) \
+                        .first()
+                    if not existing_zone:
+                        reordered_zones.append(zone)
+                        continue
+
+                    if dry_run:
+                        logger.info('Would remove zone {} and reassign to zone {}'.format(zone.id, existing_zone.id))
+                        continue
+
+                    logger.info('Removing zone {} and reassigning to zone {}…'.format(zone.id, existing_zone.id))
+                    zone.transcriptions.all().update(zone_id=existing_zone.id)
+                    zone.elements.all().update(zone_id=existing_zone.id)
+                    zone.regions.all().update(zone_id=existing_zone.id)
+                    zone.delete()
+                    removed += 1
+                    total -= 1
 
-            Zone.objects.bulk_update(reordered_zones, ['polygon'], batch_size=update_batch_size)
+            if not dry_run:
+                Zone.objects.bulk_update(reordered_zones, ['polygon'], batch_size=update_batch_size)
             reordered += len(reordered_zones)
             completed += batch.count()