fix: FIT-1078: set task state correctly on import

label_studio/core/settings/base.py

@@ -933,6 +933,7 @@ def collect_versions_dummy(**kwargs):
 # Base FSM (Finite State Machine) Configuration for Label Studio
 FSM_CACHE_TTL = 300  # Cache TTL in seconds (5 minutes)
 FSM_SYNC_PROJECT_STATE = 'fsm.project_transitions.sync_project_state'
+FSM_INFERENCE_FUNCTION = 'fsm.state_inference._get_or_infer_state'
 
 # Used for async migrations. In LSE this is set to a real queue name, including here so we
 # can use settings.SERVICE_QUEUE_NAME in async migrations in LSO

label_studio/fsm/README.md

@@ -309,4 +309,5 @@ When contributing:
 - Keep framework code generic and reusable
 - Add performance tests for UUID7 optimizations
 - Document extension points and customization options
-- Ensure extensibility is preserved
+- Ensure extensibility is preserved
+- When adding pytests, use and extend reusable helper functions in `fsm/tests/helpers.py` when appropriate

label_studio/fsm/functions.py

@@ -4,9 +4,12 @@
 This module contains reusable functions for FSM state management that are
 used across different parts of the codebase.
 """
-
 import logging
 
+from core.current_request import CurrentContext
+from fsm.state_inference import get_or_infer_state
+from fsm.utils import get_or_initialize_state
+
 logger = logging.getLogger(__name__)
 
 
@@ -32,7 +35,6 @@ def backfill_fsm_states_for_tasks(storage_id, tasks_created, link_class):
         return
 
     try:
-        from lse_fsm.state_inference import backfill_state_for_entity
         from tasks.models import Task
 
         # Get tasks created in this sync
@@ -48,12 +50,11 @@ def backfill_fsm_states_for_tasks(storage_id, tasks_created, link_class):
 
         # Backfill initial CREATED state for each task
         for task in tasks:
-            backfill_state_for_entity(task, 'task', create_record=True)
+
+            inferred_state = get_or_infer_state(task)
+            get_or_initialize_state(task, user=CurrentContext.get_user(), inferred_state=inferred_state)
 
         logger.info(f'Storage sync: FSM states created for {len(task_ids)} tasks')
-    except ImportError:
-        # LSE not available (OSS), skip FSM sync
-        logger.debug('LSE not available, skipping FSM state backfill for storage sync')
     except Exception as e:
         # Don't fail storage sync if FSM sync fails
         logger.error(f'FSM sync after storage sync failed: {e}', exc_info=True)

label_studio/fsm/project_transitions.py

@@ -11,9 +11,10 @@
 from django.conf import settings
 from fsm.registry import register_state_transition
 from fsm.state_choices import ProjectStateChoices
+from fsm.state_inference import get_or_infer_state
 from fsm.state_manager import StateManager
 from fsm.transitions import ModelChangeTransition, TransitionContext
-from fsm.utils import get_or_initialize_state, infer_entity_state_from_data
+from fsm.utils import get_or_initialize_state
 
 
 @register_state_transition('project', 'project_created', triggers_on_create=True, triggers_on_update=False)
@@ -135,7 +136,7 @@ def transition(self, context: TransitionContext) -> Dict[str, Any]:
 
 def sync_project_state(project, user=None, reason=None, context_data=None):
     current_state = StateManager.get_current_state_value(project)
-    inferred_state = infer_entity_state_from_data(project)
+    inferred_state = get_or_infer_state(project)
 
     if current_state is None:
         get_or_initialize_state(project, user=user, inferred_state=inferred_state)

label_studio/fsm/state_inference.py

@@ -0,0 +1,77 @@
+import logging
+from typing import Optional
+
+from core.utils.common import load_func
+from django.conf import settings
+
+logger = logging.getLogger(__name__)
+
+
+def _get_or_infer_state(entity) -> Optional[str]:
+    """
+    Infer what the FSM state should be based on entity's current data.
+
+    This is used for "cold start" scenarios where entities exist in the database
+    but don't have FSM state records yet (e.g., after FSM deployment to production
+    with pre-existing data).
+
+    Args:
+        entity: The entity to infer state for (Task, Project, or Annotation)
+
+    Returns:
+        Inferred state value, or None if entity type not supported
+
+    Examples:
+        >>> task = Task.objects.get(id=123)
+        >>> task.is_labeled = True
+        >>> _get_or_infer_state(task)
+        'COMPLETED'
+
+        >>> project = Project.objects.get(id=456)
+        >>> _get_or_infer_state(project)
+        'CREATED'
+    """
+    from fsm.state_choices import AnnotationStateChoices, ProjectStateChoices, TaskStateChoices
+
+    entity_type = entity._meta.model_name.lower()
+
+    if entity_type == 'task':
+        # Task state depends on whether it has been labeled
+        return TaskStateChoices.COMPLETED if entity.is_labeled else TaskStateChoices.CREATED
+    elif entity_type == 'project':
+        # Project state depends on task completion
+        # If no tasks exist, project is CREATED
+        # If any tasks are completed, project is at least IN_PROGRESS
+        # If all tasks are completed, project is COMPLETED
+        tasks = entity.tasks.all()
+        if not tasks.exists():
+            return ProjectStateChoices.CREATED
+
+        # Count labeled tasks to determine project state
+        total_tasks = tasks.count()
+        labeled_tasks = tasks.filter(is_labeled=True).count()
+
+        if labeled_tasks == 0:
+            return ProjectStateChoices.CREATED
+        elif labeled_tasks == total_tasks:
+            return ProjectStateChoices.COMPLETED
+        else:
+            return ProjectStateChoices.IN_PROGRESS
+    elif entity_type == 'annotation':
+        # Annotations are SUBMITTED when created
+        return AnnotationStateChoices.SUBMITTED
+    else:
+        logger.warning(
+            f'Cannot infer state for unknown entity type: {entity_type}',
+            extra={
+                'event': 'fsm.infer_state_unknown_type',
+                'entity_type': entity_type,
+                'entity_id': entity.pk,
+            },
+        )
+        return None
+
+
+def get_or_infer_state(entity) -> Optional[str]:
+    func = load_func(settings.FSM_INFERENCE_FUNCTION)
+    return func(entity)

label_studio/fsm/tests/conftest.py

@@ -12,9 +12,25 @@
 from fsm.registry import state_choices_registry, state_model_registry, transition_registry
 from fsm.state_manager import StateManager
 
+from label_studio.tests import conftest as ls_tests_conftest
+
 logger = logging.getLogger(__name__)
 
 
+# Re-export core fixtures from main LS test suite so FSM tests behave like OSS tests.
+# NOTE: We alias the same underlying functions so pytest registers them as fixtures
+# in this module as well (including their autouse behavior).
+django_live_url = ls_tests_conftest.get_server_url
+business_client = ls_tests_conftest.business_client
+
+# Storage-related fixtures to mock cloud providers for import storages
+aws_credentials = ls_tests_conftest.aws_credentials
+s3 = ls_tests_conftest.s3
+s3_with_images = ls_tests_conftest.s3_with_images
+gcs_client = ls_tests_conftest.gcs_client
+azure_client = ls_tests_conftest.azure_client
+
+
 @pytest.fixture(autouse=True, scope='function')
 def fsm_test_isolation():
     """