Merge branch 'create-transcriptions-list-elements' into 'master'

Add a parameter to serialize transcription elements in CreateElementTranscriptions

See merge request !863

arkindex/documents/api/ml.py

@@ -13,7 +13,7 @@ from arkindex.documents.serializers.ml import (
     ClassificationsSerializer, ClassificationCreateSerializer, ClassificationSerializer,
     TranscriptionsSerializer, TranscriptionSerializer, TranscriptionCreateSerializer,
     DataSourceStatsSerializer, ClassificationsSelectionSerializer, ClassificationMode,
-    CountMLClassSerializer, MLClassSerializer, ElementTranscriptionsBulkSerializer,
+    CountMLClassSerializer, MLClassSerializer, ElementTranscriptionsBulkSerializer, AnnotatedElementSerializer
 )
 from arkindex.images.importer import build_transcriptions, save_transcriptions
 from arkindex.images.models import Zone
@@ -126,6 +126,35 @@ class ElementTranscriptionsBulk(CreateAPIView):
         'description': 'Create multiple sub elements with their transcriptions on a common parent in one transaction',
         'operationId': 'CreateElementTranscriptions',
         'tags': ['transcriptions'],
+        'responses': {
+            '201': {
+                'description': 'Created',
+                'content': {
+                    'application/json': {
+                        'examples': [{
+                            'summary': 'return_elements = false',
+                            'value': {
+                                'element_type': 'string',
+                                'transcription_type': 'string',
+                                'source': 'string',
+                                'transcriptions': {
+                                    'polygon': [[]],
+                                    'text': 'string',
+                                    'score': 0
+                                },
+                                'return_elements': False
+                            }
+                        }, {
+                            'summary': 'return_elements = true',
+                            'value': [{
+                                'id': 'string',
+                                'created': True
+                            }]
+                        }]
+                    }
+                }
+            }
+        }
     }
 
     def get_object(self):
@@ -144,6 +173,20 @@ class ElementTranscriptionsBulk(CreateAPIView):
             context['element'] = self.get_object()
         return context
 
+    def create(self, request, *args, **kwargs):
+        serializer = self.get_serializer(data=request.data)
+        serializer.is_valid(raise_exception=True)
+        annotations = self.perform_create(serializer)
+        headers = self.get_success_headers(serializer.data)
+
+        if serializer.validated_data['return_elements']:
+            # Re-serialize the annotated sub-elements
+            data = AnnotatedElementSerializer(annotations, many=True).data
+        else:
+            data = serializer.data
+
+        return Response(data, status=status.HTTP_201_CREATED, headers=headers)
+
     @transaction.atomic
     def perform_create(self, serializer):
         elt_type = serializer.validated_data['element_type']
@@ -170,6 +213,7 @@ class ElementTranscriptionsBulk(CreateAPIView):
                     image_id=image.id,
                     polygon=annotation['polygon'],
                 )
+                existing_zones[annotation['polygon']] = annotation['zone']
                 missing_zones.append(annotation['zone'])
         Zone.objects.bulk_create(missing_zones)
 
@@ -198,6 +242,9 @@ class ElementTranscriptionsBulk(CreateAPIView):
                     type=elt_type,
                     name=next_path_ordering + 1
                 )
+                # Specify the annotated element has been created
+                annotation['created'] = True
+                children[annotation['zone'].id] = annotation['element']
                 missing_elements.append(annotation['element'])
                 for parent_path in paths:
                     new_path = parent_path.path + [self.element.id]
@@ -229,6 +276,9 @@ class ElementTranscriptionsBulk(CreateAPIView):
         # Index in ES
         reindex_start(element=self.element, transcriptions=True, elements=True)
 
+        # Return ordered annotations
+        return annotations
+
 
 class TranscriptionBulk(CreateAPIView, UpdateAPIView):
     '''

arkindex/documents/serializers/ml.py

@@ -328,6 +328,14 @@ class ElementTranscriptionsBulkSerializer(serializers.Serializer):
         allow_empty=False,
         help_text='A list of transcriptions with their text, score and element zone'
     )
+    return_elements = serializers.BooleanField(
+        default=False,
+        help_text=(
+            'Return a list of annotations containing an `element` field with basic element informations '
+            'and a `created` field specifying if the element has been created for the annotation. '
+            'Elements are ordered corresponding to the `transcriptions` field and may be duplicated.'
+        )
+    )
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
@@ -342,6 +350,16 @@ class ElementTranscriptionsBulkSerializer(serializers.Serializer):
         return source
 
 
+class AnnotatedElementSerializer(serializers.Serializer):
+    """
+    Returns annotated element IDs specifying if they have been created
+    to hold a transcription.
+    """
+    id = serializers.UUIDField(source='element.id')
+    # Specify if the element has been created for the annotation
+    created = serializers.BooleanField(default=False)
+
+
 class ClassificationBulkSerializer(serializers.Serializer):
     """
     Single classification serializer for bulk insertion

arkindex/documents/tests/test_bulk_element_transcriptions.py

@@ -273,3 +273,64 @@ class TestBulkElementTranscriptions(FixtureAPITestCase):
                 'score': ['This field is required.']
             }]
         })
+
+    @patch('arkindex.project.serializer_fields.MLTool.get')
+    @override_settings(ARKINDEX_FEATURES={'search': False})
+    def test_bulk_transcriptions_return_ordered_elements(self, ml_get_mock):
+        """
+        return_elements field allows to serialize an ordered list of sub-elements holding the transcriptions
+        Returned elements may be duplicated in case the same zone is used for multiple transcriptions
+        """
+        ml_get_mock.return_value.type = self.src.type
+        ml_get_mock.return_value.slug = self.src.slug
+        ml_get_mock.return_value.name = self.src.name
+        ml_get_mock.return_value.version = self.src.revision
+        self.src.internal = True
+        self.src.save()
+
+        transcriptions = [
+            ([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'stuck', 0.13),
+            ([[13, 37], [133, 37], [133, 137], [13, 137], [13, 37]], 'stick', 0.33),
+            ([[24, 42], [64, 42], [64, 142], [24, 142], [24, 42]], 'stock', 0.37),
+            # The existing line has the smallest ordering
+            (self.line.zone.polygon, 'stack', 0.42),
+        ]
+        data = {
+            'element_type': 'text_line',
+            'transcription_type': 'line',
+            'source': self.src.slug,
+            'transcriptions': [{
+                'polygon': poly,
+                'text': text,
+                'score': score
+            } for poly, text, score in transcriptions],
+            'return_elements': True
+        }
+
+        self.client.force_login(self.internal_user)
+        with self.assertNumQueries(19):
+            response = self.client.post(
+                reverse('api:element-transcriptions-bulk', kwargs={'pk': self.page.id}),
+                format='json',
+                data=data
+            )
+        self.assertEqual(response.status_code, status.HTTP_201_CREATED)
+
+        created_elts = Element.objects.get_descending(self.page.id).exclude(id=self.line.id)
+        self.assertEqual(created_elts.count(), 2)
+
+        first_elt = created_elts.get(transcriptions__text='stuck')
+        second_elt = created_elts.get(transcriptions__text='stock')
+        self.assertListEqual(response.json(), [{
+            'id': str(first_elt.id),
+            'created': True
+        }, {
+            'id': str(first_elt.id),
+            'created': False
+        }, {
+            'id': str(second_elt.id),
+            'created': True
+        }, {
+            'id': str(self.line.id),
+            'created': False
+        }])