Merge branch 'reorder-dedupe' into 'master'

Deduplicate zones in reorder_polygons

See merge request !606

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()