diff --git a/arkindex/documents/api/elements.py b/arkindex/documents/api/elements.py
index fe38b288b363e1fb2f00debf5209cb9fbef16d2c..e789d173a13c381369f082125fb72907f47d719d 100644
--- a/arkindex/documents/api/elements.py
+++ b/arkindex/documents/api/elements.py
@@ -31,7 +31,7 @@ from rest_framework.generics import (
from rest_framework.mixins import DestroyModelMixin
from rest_framework.response import Response
-from arkindex.dataimport.models import WorkerVersion
+from arkindex.dataimport.models import WorkerActivity, WorkerActivityState, WorkerVersion
from arkindex.documents.models import (
AllowedMetaData,
Classification,
@@ -54,6 +54,7 @@ from arkindex.documents.serializers.elements import (
ElementSlimSerializer,
ElementTypeSerializer,
MetaDataUpdateSerializer,
+ WorkerStatisticsSerializer,
)
from arkindex.documents.serializers.light import CorpusAllowedMetaDataSerializer, ElementTypeLightSerializer
from arkindex.documents.serializers.ml import ElementTranscriptionSerializer
@@ -1410,3 +1411,38 @@ class WorkerResultsDestroy(CorpusACLMixin, DestroyAPIView):
)
return Response(status=status.HTTP_204_NO_CONTENT)
+
+
+@extend_schema_view(
+ get=extend_schema(
+ operation_id='RetriveWorkersActivity',
+ tags=['elements']
+ )
+)
+class WorkersActivity(CorpusACLMixin, ListAPIView):
+ """
+ Retrieve corpus wise statistics about the activity of a single worker version
+ """
+ permission_classes = (IsVerified, )
+ serializer_class = WorkerStatisticsSerializer
+ pagination_class = None
+ queryset = WorkerActivity.objects.none()
+
+ def list(self, request, *args, **kwargs):
+ corpus = self.get_corpus(self.kwargs['corpus'], role=Role.Admin)
+
+ # Retrieve the distribution of activities on this corpus grouped by worker version
+ stats = WorkerActivity.objects \
+ .filter(element_id__in=corpus.elements.values('id')) \
+ .values('worker_version_id') \
+ .annotate(
+ **{
+ state.value: Count('id', filter=Q(state=state.value))
+ for state in WorkerActivityState
+ }
+ )
+
+ return Response(
+ status=status.HTTP_200_OK,
+ data=WorkerStatisticsSerializer(stats, many=True).data
+ )
diff --git a/arkindex/documents/serializers/elements.py b/arkindex/documents/serializers/elements.py
index 1acbaafbecfe8908a45288eb815cd08e46a61377..1cddedda6033656c4a9184d9f0f9b2d6cf5c5b2a 100644
--- a/arkindex/documents/serializers/elements.py
+++ b/arkindex/documents/serializers/elements.py
@@ -669,3 +669,14 @@ class ElementBulkSerializer(serializers.Serializer):
element['type'] = type_ids[element['type']]
return data
+
+
+class WorkerStatisticsSerializer(serializers.Serializer):
+ """
+ Serialize activity statistics of a worker version
+ """
+ worker_version_id = serializers.UUIDField(read_only=True)
+ queued = serializers.IntegerField(read_only=True)
+ started = serializers.IntegerField(read_only=True)
+ processed = serializers.IntegerField(read_only=True)
+ error = serializers.IntegerField(read_only=True)
diff --git a/arkindex/documents/tests/test_workers_activity.py b/arkindex/documents/tests/test_workers_activity.py
new file mode 100644
index 0000000000000000000000000000000000000000..d3b61e053eba16ede895fc45272941f9b7a58fb6
--- /dev/null
+++ b/arkindex/documents/tests/test_workers_activity.py
@@ -0,0 +1,124 @@
+import itertools
+import uuid
+
+from django.urls import reverse
+from rest_framework import status
+
+from arkindex.dataimport.models import WorkerActivity, WorkerActivityState, WorkerVersion
+from arkindex.documents.models import Corpus
+from arkindex.project.tests import FixtureAPITestCase
+from arkindex.users.models import Role, User
+
+
+class TestWorkersActivity(FixtureAPITestCase):
+
+ @classmethod
+ def setUpTestData(cls):
+ super().setUpTestData()
+ cls.version_1 = WorkerVersion.objects.get(worker__slug='reco')
+ cls.version_2 = WorkerVersion.objects.get(worker__slug='dla')
+ cls.private_corpus = Corpus.objects.create(name='private', public=False)
+ cls.elts_count = cls.corpus.elements.count()
+ # Generate worker activities
+ WorkerActivity.objects.bulk_create([
+ *(
+ WorkerActivity(
+ element_id=elt.id,
+ state=state,
+ worker_version_id=cls.version_1.id
+ ) for elt, state in zip(cls.corpus.elements.all(), itertools.cycle(WorkerActivityState))
+ ), *(
+ WorkerActivity(
+ element_id=elt.id,
+ state=WorkerActivityState.Processed.value,
+ worker_version_id=cls.version_2.id
+ ) for elt in cls.corpus.elements.all()
+ )
+ ])
+
+ def test_version_stats_requires_login(self):
+ with self.assertNumQueries(0):
+ response = self.client.get(
+ reverse('api:workers-activity', kwargs={'corpus': str(self.corpus.id)})
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {'detail': 'Authentication credentials were not provided.'})
+
+ def test_version_private_corpus(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(6):
+ response = self.client.get(
+ reverse('api:workers-activity', kwargs={'corpus': str(self.private_corpus.id)})
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {'detail': 'You do not have admin access to this corpus.'})
+
+ def test_version_stats_requires_admin(self):
+ user = User.objects.create_user('user42@test.test', 'abcd')
+ self.corpus.memberships.create(user=user, level=Role.Contributor.value)
+ self.client.force_login(user)
+ with self.assertNumQueries(5):
+ response = self.client.get(
+ reverse('api:workers-activity', kwargs={'corpus': str(self.corpus.id)})
+ )
+ self.assertEqual(response.status_code, status.HTTP_403_FORBIDDEN)
+ self.assertDictEqual(response.json(), {'detail': 'You do not have admin access to this corpus.'})
+
+ def test_versions_unexisting_corpus(self):
+ self.client.force_login(self.user)
+ with self.assertNumQueries(3):
+ response = self.client.get(
+ reverse('api:workers-activity', kwargs={'corpus': str(uuid.uuid4())})
+ )
+ self.assertEqual(response.status_code, status.HTTP_404_NOT_FOUND)
+
+ def test_versions_empty(self):
+ """
+ Handle a version that has absolutely no results on the corpus
+ """
+ user = User.objects.create_user('user42@test.test', 'abcd')
+ self.private_corpus.memberships.create(user=user, level=Role.Admin.value)
+ self.client.force_login(user)
+ with self.assertNumQueries(6):
+ response = self.client.get(
+ reverse('api:workers-activity', kwargs={'corpus': str(self.private_corpus.id)})
+ )
+ self.assertEqual(response.status_code, status.HTTP_200_OK)
+ self.assertListEqual(response.json(), [])
+
+ def test_workers_activity_distributed_states(self):
+ error, processed, queued, started = [
+ WorkerActivity.objects.filter(
+ element__corpus_id=self.corpus.id,
+ worker_version_id=self.version_1.id,
+ state=state
+ ).count()
+ for state in [
+ WorkerActivityState.Error,
+ WorkerActivityState.Processed,
+ WorkerActivityState.Queued,
+ WorkerActivityState.Started
+ ]
+ ]
+ assert error > 0 and processed > 0 and queued > 0 and started > 0
+ self.client.force_login(self.user)
+ with self.assertNumQueries(7):
+ response = self.client.get(
+ reverse('api:workers-activity', kwargs={'corpus': str(self.corpus.id)})
+ )
+ self.assertEqual(response.status_code, status.HTTP_200_OK)
+ self.assertCountEqual(response.json(), [
+ {
+ 'worker_version_id': str(self.version_1.id),
+ 'queued': queued,
+ 'started': started,
+ 'processed': processed,
+ 'error': error,
+ }, {
+ 'worker_version_id': str(self.version_2.id),
+ 'queued': 0,
+ 'started': 0,
+ 'processed': self.corpus.elements.count(),
+ 'error': 0,
+ }
+ ])
diff --git a/arkindex/project/api_v1.py b/arkindex/project/api_v1.py
index 7a1fdc735ed90070dedf3e1d8362c15e69b06382..2f14c77b1d37b8ab460421ca1be09528418f69b8 100644
--- a/arkindex/project/api_v1.py
+++ b/arkindex/project/api_v1.py
@@ -49,6 +49,7 @@ from arkindex.documents.api.elements import (
ManageSelection,
MetadataEdit,
WorkerResultsDestroy,
+ WorkersActivity,
)
from arkindex.documents.api.entities import (
CorpusRoles,
@@ -137,6 +138,8 @@ api = [
path('corpus/<uuid:pk>/selection/', CorpusSelectionDestroy.as_view(), name='corpus-delete-selection'),
path('corpus/<uuid:pk>/search/', CorpusSearch.as_view(), name='corpus-search'),
path('corpus/<uuid:corpus>/workerversion/<uuid:version>/results/', WorkerResultsDestroy.as_view(), name='worker-delete-results'),
+ path('corpus/<uuid:corpus>/workers-activity/', WorkersActivity.as_view(), name='workers-activity'),
+
# Moderation
path('classifications/', ClassificationCreate.as_view(), name='classification-create'),