Support top-level elements in CreateElementTranscriptions

arkindex/documents/api/ml.py

@@ -271,7 +271,10 @@ class ElementTranscriptionsBulk(CreateAPIView):
             .only('id', 'zone_id')
         }
         # Load the paths immediately to avoid iterating over them for each element
-        paths = list(self.element.paths.all())
+        paths = list(self.element.paths.values_list('path', flat=True))
+        if not paths:
+            # Support top level elements, by adding an empty initial path to trigger loops below
+            paths = [[]]
         next_path_ordering = self.element.get_next_order(elt_type)
         for annotation in annotations:
             # Look for a direct children with the right type and zone
@@ -289,7 +292,7 @@ class ElementTranscriptionsBulk(CreateAPIView):
                 children[annotation['zone_id']] = annotation['element']
                 missing_elements.append(annotation['element'])
                 for parent_path in paths:
-                    new_path = parent_path.path + [self.element.id]
+                    new_path = parent_path + [self.element.id]
                     # Add the children to all of its parent paths
                     missing_paths.append(ElementPath(
                         element=annotation['element'],

arkindex/documents/tests/test_bulk_element_transcriptions.py

@@ -319,3 +319,68 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
         )
         self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
         self.assertEqual(response.json(), {'element': ['Cannot create transcriptions on an element without a zone.']})
+
+    @patch('arkindex.project.triggers.tasks.reindex_start.delay')
+    def test_top_level_element(self, delay_mock):
+        """
+        Create transcriptions on a top level element
+        """
+        # Create a top level page
+        top_level = self.corpus.elements.create(
+            type=self.page.type,
+            name='Top level page',
+            zone=self.page.zone,
+        )
+
+        transcriptions = [
+            ([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'Hello world !', 0.1337),
+            ([[24, 42], [64, 42], [64, 142], [24, 142], [24, 42]], 'I <3 JavaScript', 0.42),
+        ]
+        data = {
+            'element_type': 'text_line',
+            'transcription_type': 'line',
+            'worker_version': str(self.worker_version.id),
+            'transcriptions': [{
+                'polygon': poly,
+                'text': text,
+                'score': score
+            } for poly, text, score in transcriptions]
+        }
+        self.client.force_login(self.user)
+        response = self.client.post(
+            reverse('api:element-transcriptions-bulk', kwargs={'pk': top_level.id}),
+            format='json',
+            data=data
+        )
+        self.assertEqual(response.status_code, status.HTTP_201_CREATED)
+
+        created_elts = Element.objects.get_descending(top_level.id)
+        self.assertEqual(created_elts.count(), 2)
+        self.assertTrue(all(map(lambda elt: elt.zone.image == self.page.zone.image, created_elts)))
+        self.assertListEqual(
+            [
+                (elt.paths.first().ordering, elt.name, elt.zone.polygon.coords)
+                for elt in created_elts
+            ],
+            [
+                (0, '1', ((13, 37), (13, 137), (133, 137), (133, 37), (13, 37))),
+                (1, '2', ((24, 42), (24, 142), (64, 142), (64, 42), (24, 42)))
+            ]
+        )
+        self.assertCountEqual(
+            created_elts.values_list('transcriptions__type', 'transcriptions__text', 'transcriptions__source', 'transcriptions__worker_version'),
+            [
+                (TranscriptionType.Line, ('Hello world !'), None, self.worker_version.id),
+                (TranscriptionType.Line, ('I <3 JavaScript'), None, self.worker_version.id)
+            ]
+        )
+        self.assertEqual(delay_mock.call_count, 1)
+        self.assertEqual(delay_mock.call_args, call(
+            element_id=str(top_level.id),
+            corpus_id=None,
+            entity_id=None,
+            transcriptions=True,
+            elements=True,
+            entities=False,
+            drop=False,
+        ))