diff --git a/arkindex/documents/api/ml.py b/arkindex/documents/api/ml.py
index 60547b495137baabae825fe8352495d9e3346931..9980943970ecff969fcd2e7c95bce72312e3854a 100644
--- a/arkindex/documents/api/ml.py
+++ b/arkindex/documents/api/ml.py
@@ -19,13 +19,11 @@ from arkindex.documents.serializers.ml import (
CountMLClassSerializer, MLClassSerializer
)
from arkindex.documents.pagexml import PageXmlParser
-from arkindex.images.models import Zone
from arkindex.images.importer import build_transcriptions, save_transcriptions
from arkindex.project.filters import SafeSearchFilter
from arkindex.project.mixins import SelectionMixin, CorpusACLMixin
from arkindex.project.parsers import XMLParser
from arkindex.project.permissions import IsVerified, IsAdminUser
-from arkindex.project.polygon import Polygon
from arkindex.project.triggers import reindex_start, ml_results_delete
import os.path
import logging
@@ -36,6 +34,7 @@ logger = logging.getLogger(__name__)
class TranscriptionCreate(CreateAPIView):
"""
Create a single transcription on an element
+ The transcription zone is defined by the element it is attached
"""
serializer_class = TranscriptionCreateSerializer
permission_classes = (IsVerified, )
@@ -50,18 +49,10 @@ class TranscriptionCreate(CreateAPIView):
raise ValidationError("This endpoint can only import transcriptions.")
element = serializer.validated_data['element']
- # Create a zone on the page's image
- polygon = Polygon(serializer.validated_data['polygon'])
- ts_zone, _ = Zone.objects.get_or_create(
- image=element.zone.image,
- polygon=polygon,
- )
-
ts, created = Transcription.objects.get_or_create(
element=element,
source=serializer.validated_data['source'],
type=transcription_type,
- zone=ts_zone,
text=serializer.validated_data['text'],
defaults={
'score': serializer.validated_data['score']
diff --git a/arkindex/documents/serializers/ml.py b/arkindex/documents/serializers/ml.py
index b49d0e9e3e7930200a90fcbde82a05db09f47b49..277832fcb3c8ab9745274fbb89e25baaa728e58c 100644
--- a/arkindex/documents/serializers/ml.py
+++ b/arkindex/documents/serializers/ml.py
@@ -176,7 +176,7 @@ class TranscriptionCreateSerializer(serializers.Serializer):
queryset=Element.objects.filter(zone__isnull=False),
)
source = DataSourceSlugField(tool_type=MLToolType.Recognizer)
- polygon = PolygonField()
+ polygon = PolygonField(read_only=True)
text = serializers.CharField()
score = serializers.FloatField(min_value=0, max_value=1)
type = EnumField(TranscriptionType)
diff --git a/arkindex/documents/tests/test_transcription_create.py b/arkindex/documents/tests/test_transcription_create.py
index 998f8fab541edde17b8c122c29850a0756e27d8f..e6bfc61f9e0281952c2504f792375e07be10664b 100644
--- a/arkindex/documents/tests/test_transcription_create.py
+++ b/arkindex/documents/tests/test_transcription_create.py
@@ -38,14 +38,13 @@ class TestTranscriptionCreate(FixtureAPITestCase):
"type": "word",
"element": str(self.page.id),
"source": self.src.slug,
- "polygon": [(0, 0), (100, 0), (100, 100), (0, 100), (0, 0)],
"text": "NEKUDOTAYIM",
"score": 0.83,
})
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
new_ts = Transcription.objects.get(text="NEKUDOTAYIM", type=TranscriptionType.Word)
- self.assertEqual(new_ts.zone.polygon, Polygon.from_coords(0, 0, 100, 100))
+ self.assertEqual(new_ts.zone, None)
self.assertEqual(new_ts.score, 0.83)
self.assertEqual(new_ts.source, self.src)
self.assertTrue(self.page.transcriptions.filter(pk=new_ts.id).exists())
@@ -62,33 +61,25 @@ class TestTranscriptionCreate(FixtureAPITestCase):
})
@patch('arkindex.project.triggers.get_channel_layer')
- def test_unique_zone(self, get_layer_mock):
+ def test_create_transcription_takes_no_zone(self, get_layer_mock):
"""
- Checks the view reuses zones when available
+ TranscriptionCreate endpoint do not create any zone for the transcription
"""
get_layer_mock.return_value.send = AsyncMock()
+
self.client.force_login(self.user)
- ts = Transcription.objects.get(zone__image__path='img1', text="PARIS")
response = self.client.post(reverse('api:transcription-create'), format='json', data={
- "type": "word",
+ "type": "page",
"element": str(self.page.id),
"source": self.src.slug,
- "polygon": ts.zone.polygon.serialize(),
- "text": "GLOUBIBOULGA",
- "score": 0.8,
+ "polygon": [(0, 0), (42, 0), (42, 42), (0, 42), (0, 0)],
+ "text": "SQUARE",
+ "score": 0.42,
})
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
- self.assertEqual(Transcription.objects.get(text="GLOUBIBOULGA").zone.id, ts.zone.id)
- get_layer_mock().send.assert_called_once_with('reindex', {
- 'type': 'reindex.start',
- 'element': str(self.page.id),
- 'corpus': None,
- 'entity': None,
- 'transcriptions': True,
- 'elements': True,
- 'entities': False,
- 'drop': False,
- })
+
+ new_ts = Transcription.objects.get(text="SQUARE", type=TranscriptionType.Page)
+ self.assertEqual(new_ts.zone, None)
@patch('arkindex.project.triggers.get_channel_layer')
def test_update_transcription(self, get_layer_mock):
@@ -103,7 +94,6 @@ class TestTranscriptionCreate(FixtureAPITestCase):
"type": "word",
"element": str(self.page.id),
"source": ts.source.slug,
- "polygon": ts.zone.polygon.serialize(),
"text": ts.text,
"score": 0.99,
})
@@ -123,10 +113,11 @@ class TestTranscriptionCreate(FixtureAPITestCase):
'drop': False,
})
+ @override_settings(ARKINDEX_FEATURES={'search': False})
def test_invalid_data(self):
"""
Checks the view validates data properly
- (score between 0 and 1, minimum points in polygon, non-negative coordinates, non-existent element)
+ (score between 0 and 1, non-existent element)
"""
self.client.force_login(self.user)
@@ -134,31 +125,38 @@ class TestTranscriptionCreate(FixtureAPITestCase):
"type": "word",
"element": str(self.page.id),
"source": self.src.slug,
- "polygon": [(0, 0), (0, 100), (100, 100), (100, 0), (0, 0)],
"text": "NEKUDOTAYIM",
"score": 0.83,
}
- # Negative score
- post_data['score'] = -1
+ # Assert data is valid
response = self.client.post(reverse('api:transcription-create'), format='json', data=post_data)
+ self.assertEqual(response.status_code, status.HTTP_201_CREATED)
+
+ # Negative score
+ data = {**post_data, 'score': -1}
+ response = self.client.post(reverse('api:transcription-create'), format='json', data=data)
self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+ self.assertDictEqual(response.json(), {
+ 'score': ['Ensure this value is greater than or equal to 0.']
+ })
# Score over 100%
- post_data['score'] = 2
- response = self.client.post(reverse('api:transcription-create'), format='json', data=post_data)
+ data = {**post_data, 'score': 2}
+ response = self.client.post(reverse('api:transcription-create'), format='json', data=data)
self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+ self.assertDictEqual(response.json(), {
+ 'score': ['Ensure this value is less than or equal to 1.']
+ })
# Missing element
- post_data['score'] = 0.83
- post_data['element'] = str(uuid.uuid4())
- response = self.client.post(reverse('api:transcription-create'), format='json', data=post_data)
- self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
-
- # Not enough polygon points
- post_data['polygon'] = [(0, 0), (100, 100)]
- response = self.client.post(reverse('api:transcription-create'), format='json', data=post_data)
+ wrong_id = str(uuid.uuid4())
+ data = {**post_data, 'element': wrong_id}
+ response = self.client.post(reverse('api:transcription-create'), format='json', data=data)
self.assertEqual(response.status_code, status.HTTP_400_BAD_REQUEST)
+ self.assertDictEqual(response.json(), {
+ 'element': [f'Invalid pk "{wrong_id}" - object does not exist.']
+ })
@patch('arkindex.project.serializer_fields.MLTool.get')
@patch('arkindex.project.triggers.get_channel_layer')
@@ -410,6 +408,40 @@ class TestTranscriptionCreate(FixtureAPITestCase):
self.assertEqual(response.status_code, status.HTTP_201_CREATED)
self.assertFalse(get_layer_mock().send.called)
+ @patch('arkindex.project.serializer_fields.MLTool.get')
+ @patch('arkindex.project.triggers.get_channel_layer')
+ @override_settings(ARKINDEX_FEATURES={'search': False})
+ def test_unique_zone(self, get_layer_mock, ml_get_mock):
+ """
+ Checks the bulk view reuses zones when available
+ """
+ ml_get_mock.return_value.type = self.src.type
+ ml_get_mock.return_value.slug = 'new_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()
+
+ self.assertEqual(self.page.transcriptions.count(), 4)
+ existing_ts = self.page.transcriptions.get(zone__image__path='img1', text="PARIS")
+ data = {
+ "parent": str(self.page.id),
+ "recognizer": self.src.slug,
+ "transcriptions": [{
+ "type": "word",
+ "text": "PARISH",
+ "score": 0.999,
+ "polygon": existing_ts.zone.polygon.serialize(),
+ }]
+ }
+ self.client.force_login(self.user)
+ response = self.client.put(reverse('api:transcription-bulk'), format='json', data=data)
+ self.assertEqual(response.status_code, status.HTTP_201_CREATED)
+
+ # A new transcription has been added with the same zone
+ self.assertEqual(self.page.transcriptions.count(), 5)
+ self.assertEqual(Transcription.objects.get(text="PARISH").zone.id, existing_ts.zone.id)
+
@patch('arkindex.project.triggers.get_channel_layer')
@override_settings(ARKINDEX_FEATURES={'search': False})
def test_create_transcription_no_search(self, get_layer_mock):