feat: LEAP-648: Reset and recalculate project summary

label_studio/core/all_urls.json

@@ -179,6 +179,12 @@
         "name": "projects:api:project-summary",
         "decorators": ""
     },
+    {
+        "url": "/api/projects/<int:pk>/summary/reset/",
+        "module": "projects.api.ProjectSummaryResetAPI",
+        "name": "projects:api:project-summary-reset",
+        "decorators": ""
+    },
     {
         "url": "/api/projects/<int:pk>/imports/<int:import_pk>/",
         "module": "projects.api.ProjectImportAPI",

label_studio/projects/api.py

@@ -9,6 +9,7 @@
 from core.label_config import config_essential_data_has_changed
 from core.mixins import GetParentObjectMixin
 from core.permissions import ViewClassPermission, all_permissions
+from core.redis import start_job_async_or_sync
 from core.utils.common import paginator, paginator_help, temporary_disconnect_all_signals
 from core.utils.exceptions import LabelStudioDatabaseException, ProjectExistException
 from core.utils.io import find_dir, find_file, read_yaml
@@ -23,6 +24,7 @@
 from drf_yasg.utils import swagger_auto_schema
 from projects.functions.next_task import get_next_task
 from projects.functions.stream_history import get_label_stream_history
+from projects.functions.utils import recalculate_created_annotations_and_labels_from_scratch
 from projects.models import Project, ProjectImport, ProjectManager, ProjectReimport, ProjectSummary
 from projects.serializers import (
     GetFieldsSerializer,
@@ -397,6 +399,32 @@ def get(self, *args, **kwargs):
         return super(ProjectSummaryAPI, self).get(*args, **kwargs)
 
 
+class ProjectSummaryResetAPI(GetParentObjectMixin, generics.CreateAPIView):
+    """This API is useful when we need to reset project.summary.created_labels and created_labels_drafts
+    and recalculate them from scratch. It's hard to correctly follow all changes in annotation region
+    labels and these fields aren't calculated properly after some time. Label config changes are not allowed
+    when these changes touch any labels from these created_labels* dictionaries.
+    """
+
+    parser_classes = (JSONParser,)
+    parent_queryset = Project.objects.all()
+    permission_required = ViewClassPermission(
+        POST=all_permissions.projects_change,
+    )
+
+    @swagger_auto_schema(auto_schema=None)
+    def post(self, *args, **kwargs):
+        project = self.get_parent_object()
+        summary = project.summary
+        start_job_async_or_sync(
+            recalculate_created_annotations_and_labels_from_scratch,
+            project,
+            summary,
+            organization_id=self.request.user.active_organization.id,
+        )
+        return Response(status=status.HTTP_200_OK)
+
+
 @method_decorator(
     name='get',
     decorator=swagger_auto_schema(

label_studio/projects/functions/utils.py

@@ -1,4 +1,13 @@
-from tasks.models import Task
+from logging import getLogger
+from typing import TYPE_CHECKING
+
+from tasks.models import AnnotationDraft, Task
+
+logger = getLogger(__name__)
+
+
+if TYPE_CHECKING:
+    from projects.models import Project, ProjectSummary
 
 
 def make_queryset_from_iterable(tasks_list):
@@ -26,3 +35,29 @@ def make_queryset_from_iterable(tasks_list):
                 raise ValueError(f'Unknown object type: {str(task)}')
         queryset = Task.objects.filter(id__in=ids)
     return queryset
+
+
+def recalculate_created_annotations_and_labels_from_scratch(
+    project: 'Project', summary: 'ProjectSummary', organization_id: int
+) -> None:
+    """Recalculate created_labels, created_annotations and created_labels_drafts from scratch
+
+    :param project: Project
+    :param summary: ProjectSummary
+    :param organization_id: Organization.id, it is required for django-rq displaying on admin page
+    """
+    logger.info(f'Reset cache started for project {project.id} and organization {organization_id}')
+
+    summary.created_labels, summary.created_annotations = {}, {}
+    summary.update_created_annotations_and_labels(project.annotations.all())
+
+    summary.created_labels_drafts = {}
+    drafts = AnnotationDraft.objects.filter(task__project=project)
+    summary.update_created_labels_drafts(drafts)
+
+    logger.info(
+        f'Reset cache finished for project {project.id} and organization {organization_id}:\n'
+        f'created_annotations = {summary.created_annotations}\n'
+        f'created_labels = {summary.created_labels}\n'
+        f'created_labels_drafts = {summary.created_labels_drafts}'
+    )

label_studio/projects/models.py

@@ -1200,37 +1200,43 @@ def update_created_annotations_and_labels(self, annotations):
         self.save(update_fields=['created_annotations', 'created_labels'])
 
     def remove_created_annotations_and_labels(self, annotations):
-        created_annotations = dict(self.created_annotations)
-        labels = dict(self.created_labels)
-        for annotation in annotations:
-            results = get_attr_or_item(annotation, 'result') or []
-            if not isinstance(results, list):
-                continue
+        # we are going to remove all annotations, so we'll reset the corresponding fields on the summary
+        remove_all_annotations = self.project.annotations.count() == len(annotations)
+        created_annotations, created_labels = (
+            ({}, {}) if remove_all_annotations else (dict(self.created_annotations), dict(self.created_labels))
+        )
 
-            for result in results:
-                # reduce annotation counters
-                key = self._get_annotation_key(result)
-                if key in created_annotations:
-                    created_annotations[key] -= 1
-                    if created_annotations[key] == 0:
-                        created_annotations.pop(key)
-
-                # reduce labels counters
-                from_name = result.get('from_name', None)
-                if from_name not in labels:
+        if not remove_all_annotations:
+            for annotation in annotations:
+                results = get_attr_or_item(annotation, 'result') or []
+                if not isinstance(results, list):
                     continue
-                for label in self._get_labels(result):
-                    label = str(label)
-                    if label in labels[from_name]:
-                        labels[from_name][label] -= 1
-                        if labels[from_name][label] == 0:
-                            labels[from_name].pop(label)
-                if not labels[from_name]:
-                    labels.pop(from_name)
+
+                for result in results:
+                    # reduce annotation counters
+                    key = self._get_annotation_key(result)
+                    if key in created_annotations:
+                        created_annotations[key] -= 1
+                        if created_annotations[key] == 0:
+                            created_annotations.pop(key)
+
+                    # reduce labels counters
+                    from_name = result.get('from_name', None)
+                    if from_name not in created_labels:
+                        continue
+                    for label in self._get_labels(result):
+                        label = str(label)
+                        if label in created_labels[from_name]:
+                            created_labels[from_name][label] -= 1
+                            if created_labels[from_name][label] == 0:
+                                created_labels[from_name].pop(label)
+                    if not created_labels[from_name]:
+                        created_labels.pop(from_name)
+
         logger.debug(f'summary.created_annotations = {created_annotations}')
-        logger.debug(f'summary.created_labels = {labels}')
+        logger.debug(f'summary.created_labels = {created_labels}')
         self.created_annotations = created_annotations
-        self.created_labels = labels
+        self.created_labels = created_labels
         self.save(update_fields=['created_annotations', 'created_labels'])
 
     def update_created_labels_drafts(self, drafts):
@@ -1257,25 +1263,29 @@ def update_created_labels_drafts(self, drafts):
         self.save(update_fields=['created_labels_drafts'])
 
     def remove_created_drafts_and_labels(self, drafts):
-        labels = dict(self.created_labels_drafts)
-        for draft in drafts:
-            results = get_attr_or_item(draft, 'result') or []
-            if not isinstance(results, list):
-                continue
-
-            for result in results:
-                # reduce labels counters
-                from_name = result.get('from_name', None)
-                if from_name not in labels:
+        # we are going to remove all drafts, so we'll reset the corresponding field on the summary
+        remove_all_drafts = AnnotationDraft.objects.filter(task__project=self.project).count() == len(drafts)
+        labels = {} if remove_all_drafts else dict(self.created_labels_drafts)
+
+        if not remove_all_drafts:
+            for draft in drafts:
+                results = get_attr_or_item(draft, 'result') or []
+                if not isinstance(results, list):
                     continue
-                for label in self._get_labels(result):
-                    label = str(label)
-                    if label in labels[from_name]:
-                        labels[from_name][label] -= 1
-                        if labels[from_name][label] == 0:
-                            labels[from_name].pop(label)
-                if not labels[from_name]:
-                    labels.pop(from_name)
+
+                for result in results:
+                    # reduce labels counters
+                    from_name = result.get('from_name', None)
+                    if from_name not in labels:
+                        continue
+                    for label in self._get_labels(result):
+                        label = str(label)
+                        if label in labels[from_name]:
+                            labels[from_name][label] -= 1
+                            if labels[from_name][label] == 0:
+                                labels[from_name].pop(label)
+                    if not labels[from_name]:
+                        labels.pop(from_name)
         logger.debug(f'summary.created_labels_drafts = {labels}')
         self.created_labels_drafts = labels
         self.save(update_fields=['created_labels_drafts'])

label_studio/projects/urls.py

@@ -29,6 +29,12 @@
     path('<int:pk>/validate/', api.ProjectLabelConfigValidateAPI.as_view(), name='project-label-config-validate'),
     # Project summary
     path('<int:pk>/summary/', api.ProjectSummaryAPI.as_view(), name='project-summary'),
+    # Project summary
+    path(
+        '<int:pk>/summary/reset/',
+        api.ProjectSummaryResetAPI.as_view(),
+        name='project-summary-reset',
+    ),
     # Project import
     path('<int:pk>/imports/<int:import_pk>/', api.ProjectImportAPI.as_view(), name='project-imports'),
     # Project reimport

label_studio/tests/create_project_and_import_data.tavern.yml

@@ -32,7 +32,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -84,7 +84,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -136,7 +136,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -188,7 +188,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -240,7 +240,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -292,7 +292,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -344,7 +344,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"
@@ -396,7 +396,7 @@ stages:
       content-type: application/json
     json:
       data:
-        image: "https://develop.dev.heartex.com/static/samples/sample.jpg"
+        image: "https://heartex.com/static/samples/sample.jpg"
         options:
           - value: "a"
           - value: "b"