Merge branch 'develop' into 'fb-optic-1178/memory-leak'

Workflow run: https://github.com/HumanSignal/label-studio/actions/runs/12320284871

.github/workflows/tests.yml

@@ -179,7 +179,7 @@ jobs:
       fail-fast: false
       matrix:
         python-version:
-          - '3.9'
+          - '3.10'
 
     # required for poetry action
     # see https://github.com/marketplace/actions/install-poetry-action#running-on-windows

label_studio/core/settings/base.py

@@ -411,7 +411,14 @@
     'django.contrib.staticfiles.finders.FileSystemFinder',
     'django.contrib.staticfiles.finders.AppDirectoriesFinder',
 )
-STATICFILES_STORAGE = 'core.storage.SkipMissedManifestStaticFilesStorage'
+STORAGES = {
+    'default': {
+        'BACKEND': 'django.core.files.storage.FileSystemStorage',
+    },
+    'staticfiles': {
+        'BACKEND': 'core.storage.SkipMissedManifestStaticFilesStorage',
+    },
+}
 
 # Sessions and CSRF
 SESSION_COOKIE_SECURE = bool(int(get_env('SESSION_COOKIE_SECURE', False)))
@@ -648,7 +655,7 @@ def collect_versions_dummy(**kwargs):
 
 if get_env('MINIO_STORAGE_ENDPOINT') and not get_bool_env('MINIO_SKIP', False):
     CLOUD_FILE_STORAGE_ENABLED = True
-    DEFAULT_FILE_STORAGE = 'storages.backends.s3boto3.S3Boto3Storage'
+    STORAGES['default']['BACKEND'] = 'storages.backends.s3boto3.S3Boto3Storage'
     AWS_STORAGE_BUCKET_NAME = get_env('MINIO_STORAGE_BUCKET_NAME')
     AWS_ACCESS_KEY_ID = get_env('MINIO_STORAGE_ACCESS_KEY')
     AWS_SECRET_ACCESS_KEY = get_env('MINIO_STORAGE_SECRET_KEY')
@@ -661,7 +668,7 @@ def collect_versions_dummy(**kwargs):
 
 if get_env('STORAGE_TYPE') == 's3':
     CLOUD_FILE_STORAGE_ENABLED = True
-    DEFAULT_FILE_STORAGE = 'core.storage.CustomS3Boto3Storage'
+    STORAGES['default']['BACKEND'] = 'core.storage.CustomS3Boto3Storage'
     if get_env('STORAGE_AWS_ACCESS_KEY_ID'):
         AWS_ACCESS_KEY_ID = get_env('STORAGE_AWS_ACCESS_KEY_ID')
     if get_env('STORAGE_AWS_SECRET_ACCESS_KEY'):
@@ -681,7 +688,7 @@ def collect_versions_dummy(**kwargs):
 
 if get_env('STORAGE_TYPE') == 'azure':
     CLOUD_FILE_STORAGE_ENABLED = True
-    DEFAULT_FILE_STORAGE = 'core.storage.CustomAzureStorage'
+    STORAGES['default']['BACKEND'] = 'core.storage.CustomAzureStorage'
     AZURE_ACCOUNT_NAME = get_env('STORAGE_AZURE_ACCOUNT_NAME')
     AZURE_ACCOUNT_KEY = get_env('STORAGE_AZURE_ACCOUNT_KEY')
     AZURE_CONTAINER = get_env('STORAGE_AZURE_CONTAINER_NAME')
@@ -690,8 +697,7 @@ def collect_versions_dummy(**kwargs):
 
 if get_env('STORAGE_TYPE') == 'gcs':
     CLOUD_FILE_STORAGE_ENABLED = True
-    # DEFAULT_FILE_STORAGE = 'storages.backends.gcloud.GoogleCloudStorage'
-    DEFAULT_FILE_STORAGE = 'core.storage.AlternativeGoogleCloudStorage'
+    STORAGES['default']['BACKEND'] = 'core.storage.AlternativeGoogleCloudStorage'
     GS_PROJECT_ID = get_env('STORAGE_GCS_PROJECT_ID')
     GS_BUCKET_NAME = get_env('STORAGE_GCS_BUCKET_NAME')
     GS_EXPIRATION = timedelta(seconds=int(get_env('STORAGE_GCS_EXPIRATION_SECS', '86400')))

label_studio/data_manager/managers.py

@@ -599,7 +599,7 @@ def annotate_annotations_results(queryset):
             )
         )
     else:
-        return queryset.annotate(annotations_results=ArrayAgg('annotations__result', distinct=True))
+        return queryset.annotate(annotations_results=ArrayAgg('annotations__result', distinct=True, default=Value([])))
 
 
 def annotate_predictions_results(queryset):
@@ -610,7 +610,7 @@ def annotate_predictions_results(queryset):
             )
         )
     else:
-        return queryset.annotate(predictions_results=ArrayAgg('predictions__result', distinct=True))
+        return queryset.annotate(predictions_results=ArrayAgg('predictions__result', distinct=True, default=Value([])))
 
 
 def annotate_annotators(queryset):
@@ -619,7 +619,7 @@ def annotate_annotators(queryset):
             annotators=Coalesce(GroupConcat('annotations__completed_by'), Value(''), output_field=models.CharField())
         )
     else:
-        return queryset.annotate(annotators=ArrayAgg('annotations__completed_by', distinct=True))
+        return queryset.annotate(annotators=ArrayAgg('annotations__completed_by', distinct=True, default=Value([])))
 
 
 def annotate_predictions_score(queryset):
@@ -653,7 +653,7 @@ def annotate_annotations_ids(queryset):
     if settings.DJANGO_DB == settings.DJANGO_DB_SQLITE:
         return queryset.annotate(annotations_ids=GroupConcat('annotations__id', output_field=models.CharField()))
     else:
-        return queryset.annotate(annotations_ids=ArrayAgg('annotations__id'))
+        return queryset.annotate(annotations_ids=ArrayAgg('annotations__id', default=Value([])))
 
 
 def annotate_predictions_model_versions(queryset):
@@ -662,7 +662,7 @@ def annotate_predictions_model_versions(queryset):
             predictions_model_versions=GroupConcat('predictions__model_version', output_field=models.CharField())
         )
     else:
-        return queryset.annotate(predictions_model_versions=ArrayAgg('predictions__model_version'))
+        return queryset.annotate(predictions_model_versions=ArrayAgg('predictions__model_version', default=Value([])))
 
 
 def annotate_avg_lead_time(queryset):

label_studio/projects/serializers.py

@@ -3,6 +3,30 @@
 import bleach
 from constants import SAFE_HTML_ATTRIBUTES, SAFE_HTML_TAGS
 from django.db.models import Q
+from label_studio_sdk.label_interface import LabelInterface
+from label_studio_sdk.label_interface.control_tags import (
+    BrushLabelsTag,
+    BrushTag,
+    ChoicesTag,
+    DateTimeTag,
+    EllipseLabelsTag,
+    EllipseTag,
+    HyperTextLabelsTag,
+    KeyPointLabelsTag,
+    KeyPointTag,
+    LabelsTag,
+    NumberTag,
+    ParagraphLabelsTag,
+    PolygonLabelsTag,
+    PolygonTag,
+    RatingTag,
+    RectangleLabelsTag,
+    RectangleTag,
+    TaxonomyTag,
+    TextAreaTag,
+    TimeSeriesLabelsTag,
+    VideoRectangleTag,
+)
 from projects.models import Project, ProjectImport, ProjectOnboarding, ProjectReimport, ProjectSummary
 from rest_flex_fields import FlexFieldsModelSerializer
 from rest_framework import serializers
@@ -66,6 +90,9 @@ class ProjectSerializer(FlexFieldsModelSerializer):
     config_has_control_tags = SerializerMethodField(
         default=None, read_only=True, help_text='Flag to detect is project ready for labeling'
     )
+    config_suitable_for_bulk_annotation = serializers.SerializerMethodField(
+        default=None, read_only=True, help_text='Flag to detect is project ready for bulk annotation'
+    )
     finished_task_number = serializers.IntegerField(default=None, read_only=True, help_text='Finished tasks')
 
     queue_total = serializers.SerializerMethodField()
@@ -82,6 +109,61 @@ def user_id(self):
     def get_config_has_control_tags(project):
         return len(project.get_parsed_config()) > 0
 
+    @staticmethod
+    def get_config_suitable_for_bulk_annotation(project):
+        li = LabelInterface(project.label_config)
+
+        # List of tags that should not be present
+        disallowed_tags = [
+            LabelsTag,
+            BrushTag,
+            BrushLabelsTag,
+            EllipseTag,
+            EllipseLabelsTag,
+            KeyPointTag,
+            KeyPointLabelsTag,
+            PolygonTag,
+            PolygonLabelsTag,
+            RectangleTag,
+            RectangleLabelsTag,
+            HyperTextLabelsTag,
+            ParagraphLabelsTag,
+            TimeSeriesLabelsTag,
+            VideoRectangleTag,
+        ]
+
+        # Return False if any disallowed tag is present
+        for tag_class in disallowed_tags:
+            if li.find_tags_by_class(tag_class):
+                return False
+
+        # Check perRegion/perItem for expanded list of tags, plus value="no" for Choices/Taxonomy
+        allowed_tags_for_checks = [ChoicesTag, TaxonomyTag, DateTimeTag, NumberTag, RatingTag, TextAreaTag]
+        for tag_class in allowed_tags_for_checks:
+            tags = li.find_tags_by_class(tag_class)
+            for tag in tags:
+                per_region = tag.attr.get('perRegion', 'false').lower() == 'true'
+                per_item = tag.attr.get('perItem', 'false').lower() == 'true'
+                if per_region or per_item:
+                    return False
+                # For ChoicesTag and TaxonomyTag, the value attribute must not be set at all
+                if tag_class in [ChoicesTag, TaxonomyTag]:
+                    if 'value' in tag.attr:
+                        return False
+
+        # For TaxonomyTag, check labeling and apiUrl
+        taxonomy_tags = li.find_tags_by_class(TaxonomyTag)
+        for tag in taxonomy_tags:
+            labeling = tag.attr.get('labeling', 'false').lower() == 'true'
+            if labeling:
+                return False
+            api_url = tag.attr.get('apiUrl', None)
+            if api_url is not None:
+                return False
+
+        # If all checks pass, return True
+        return True
+
     @staticmethod
     def get_parsed_label_config(project):
         return project.get_parsed_config()
@@ -156,6 +238,7 @@ class Meta:
             'finished_task_number',
             'queue_total',
             'queue_done',
+            'config_suitable_for_bulk_annotation',
         ]
 
     def validate_label_config(self, value):

label_studio/tests/config_validation.tavern.yml

@@ -1435,4 +1435,73 @@ stages:
     method: POST
     url: '{django_live_url}/api/projects/{pk}/validate'
   response:
-    status_code: 200
+    status_code: 200
+
+---
+test_name: check_config_suitable_for_bulk_annotation
+strict: false
+marks:
+- usefixtures:
+  - django_live_url
+stages:
+
+- id: signup
+  type: ref
+
+- name: create classification project
+  request:
+    data:
+      label_config: |
+        <View>
+          <Text name="text" value="$text"/>
+          <Choices name="sentiment" toName="text" choice="single">
+            <Choice value="Positive"/>
+            <Choice value="Neutral"/>
+            <Choice value="Negative"/>
+          </Choices>
+        </View>
+    method: POST
+    url: '{django_live_url}/api/projects'
+  response:
+    status_code: 201
+    save:
+      json:
+        classification_project_id: id
+
+- name: check classification project property
+  request:
+    method: GET
+    url: '{django_live_url}/api/projects/{classification_project_id}'
+  response:
+    status_code: 200
+    json:
+      config_suitable_for_bulk_annotation: true
+
+- name: create object detection project
+  request:
+    data:
+      label_config: |
+        <View>
+          <Image name="image" value="$image"/>
+          <RectangleLabels name="label" toName="image">
+            <Label value="Car"/>
+            <Label value="Tree"/>
+            <Label value="Person"/>
+          </RectangleLabels>
+        </View>
+    method: POST
+    url: '{django_live_url}/api/projects'
+  response:
+    status_code: 201
+    save:
+      json:
+        detection_project_id: id
+
+- name: check object detection project property
+  request:
+    method: GET
+    url: '{django_live_url}/api/projects/{detection_project_id}'
+  response:
+    status_code: 200
+    json:
+      config_suitable_for_bulk_annotation: false