Integrate validity check to update_config

Author: paulnoirel

libs/labelbox/src/labelbox/schema/workflow/__init__.py

@@ -44,8 +44,8 @@
 # Import from monolithic project_filter.py file
 from labelbox.schema.workflow.project_filter import (
     ProjectWorkflowFilter,
-    created_by,
     labeled_by,
+    created_by,  # Deprecated, use labeled_by instead
     annotation,
     dataset,
     issue_category,
@@ -95,8 +95,8 @@
     "ProjectWorkflowGraph",
     "ProjectWorkflowFilter",
     # Filter construction functions
-    "created_by",
     "labeled_by",
+    "created_by",  # Deprecated, use labeled_by instead
     "annotation",
     "sample",
     "dataset",

libs/labelbox/src/labelbox/schema/workflow/enums.py

@@ -106,7 +106,7 @@ class FilterField(str, Enum):
     """
 
     # User and creation filters
-    CreatedBy = "CreatedBy"
+    LabeledBy = "CreatedBy"  # Maps to backend CreatedBy field
 
     # Annotation and content filters
     Annotation = "Annotation"

libs/labelbox/src/labelbox/schema/workflow/nodes/logic_node.py

@@ -201,7 +201,7 @@ def remove_filter(self, filter_field: FilterField) -> "LogicNode":
 
         Args:
             filter_field: FilterField enum value specifying which filter type to remove
-                         (e.g., FilterField.CreatedBy, FilterField.Sample, FilterField.LabelingTime)
+                         (e.g., FilterField.LabeledBy, FilterField.Sample, FilterField.LabelingTime)
 
         Returns:
             LogicNode: Self for method chaining
@@ -211,7 +211,7 @@ def remove_filter(self, filter_field: FilterField) -> "LogicNode":
             >>>
             >>> # Type-safe enum approach (required)
             >>> logic.remove_filter(FilterField.Sample)
-            >>> logic.remove_filter(FilterField.CreatedBy)
+            >>> logic.remove_filter(FilterField.LabeledBy)  # Consistent with labeled_by() function
             >>> logic.remove_filter(FilterField.LabelingTime)
             >>> logic.remove_filter(FilterField.Metadata)
         """
@@ -373,7 +373,7 @@ def get_filters(self) -> "ProjectWorkflowFilter":
             >>> logic = workflow.get_node_by_id("some-logic-node-id")
             >>> user_filters = logic.get_filters()
             >>> # Add a new filter
-            >>> user_filters.append(created_by(["new-user-id"]))
+            >>> user_filters.append(labeled_by(["new-user-id"]))
             >>> # Apply the updated filters back to the node
             >>> logic.set_filters(user_filters)
         """
@@ -395,25 +395,25 @@ def add_filter(self, filter_rule: Dict[str, Any]) -> "LogicNode":
 
         Args:
             filter_rule: Filter rule from filter functions
-                        (e.g., created_by(["user_id"]), labeling_time.greater_than(300))
+                        (e.g., labeled_by(["user_id"]), labeling_time.greater_than(300))
 
         Returns:
             LogicNode: Self for method chaining
 
         Example:
-            >>> from labelbox.schema.workflow.project_filter import created_by, labeling_time, metadata, condition
+            >>> from labelbox.schema.workflow.project_filter import labeled_by, labeling_time, metadata, condition
             >>>
-            >>> logic.add_filter(created_by(["user-123"]))
+            >>> logic.add_filter(labeled_by(["user-123"]))
             >>> logic.add_filter(labeling_time.greater_than(300))
             >>> logic.add_filter(metadata([condition.contains("tag", "test")]))
-            >>> # Adding another created_by filter will replace the previous one
-            >>> logic.add_filter(created_by(["user-456"]))  # Replaces previous created_by filter
+            >>> # Adding another labeled_by filter will replace the previous one
+            >>> logic.add_filter(labeled_by(["user-456"]))  # Replaces previous labeled_by filter
         """
         # Validate that this looks like filter function output
         if not self._is_filter_function_output(filter_rule):
             raise ValueError(
                 "add_filter() only accepts output from filter functions. "
-                "Use functions like created_by(), labeling_time.greater_than(), etc."
+                "Use functions like labeled_by(), labeling_time.greater_than(), etc."
             )
 
         # Get the field name from the filter rule to check for existing filters
@@ -457,7 +457,7 @@ def _is_filter_function_output(self, filter_rule: Dict[str, Any]) -> bool:
 
         # Map backend field names to FilterField enum values
         backend_to_field = {
-            "CreatedBy": FilterField.CreatedBy,
+            "CreatedBy": FilterField.LabeledBy,  # Backend CreatedBy maps to user-facing LabeledBy
             "Annotation": FilterField.Annotation,
             "LabeledAt": FilterField.LabeledAt,
             "Sample": FilterField.Sample,

libs/labelbox/src/labelbox/schema/workflow/project_filter.py

@@ -302,10 +302,10 @@ def metadata(
     return result
 
 
-def created_by(
+def labeled_by(
     user_ids: List[str], label: Optional[str] = None
 ) -> Dict[str, Any]:
-    """Filter by users who created the labels.
+    """Filter by users who labeled the data.
 
     Args:
         user_ids: List of user IDs
@@ -320,10 +320,13 @@ def created_by(
     return result
 
 
-def labeled_by(
+def created_by(
     user_ids: List[str], label: Optional[str] = None
 ) -> Dict[str, Any]:
-    """Filter by users who labeled the data.
+    """Filter by users who created the labels.
+
+    .. deprecated:: 2.1.0
+        Use `labeled_by()` instead. This function will be removed in a future version.
 
     Args:
         user_ids: List of user IDs
@@ -332,10 +335,15 @@ def labeled_by(
     Returns:
         Dict representing the filter rule
     """
-    result: Dict[str, Any] = {"LabeledBy": user_ids}
-    if label is not None:
-        result["__label"] = label
-    return result
+    import warnings
+
+    warnings.warn(
+        "created_by() is deprecated and will be removed in a future version. "
+        "Use labeled_by() instead.",
+        DeprecationWarning,
+        stacklevel=2,
+    )
+    return labeled_by(user_ids, label)
 
 
 def annotation(
@@ -575,7 +583,7 @@ class ProjectWorkflowFilter(BaseModel):
 
     Example Usage:
         filters = ProjectWorkflowFilter([
-            created_by(["user-123"]),
+            labeled_by(["user-123"]),
             sample(20),
             labeled_at.between("2024-01-01", "2024-12-31"),
             metadata([condition.contains("tag", "test")]),
@@ -586,7 +594,7 @@ class ProjectWorkflowFilter(BaseModel):
         logic.set_filters(filters)
 
         # Or add individual filters
-        logic.add_filter(created_by(["user-123"]))
+        logic.add_filter(labeled_by(["user-123"]))
     """
 
     rules: List[Dict[str, Any]] = Field(default_factory=lambda: [])
@@ -636,7 +644,7 @@ def _validate_filter_structure(self, rule: Dict[str, Any]) -> None:
         if not isinstance(rule, dict) or not rule:
             raise ValueError(
                 "Filters must be created using filter functions. "
-                "Use functions like created_by([...]), metadata([...]), labeled_at.between(...), etc."
+                "Use functions like labeled_by([...]), metadata([...]), labeled_at.between(...), etc."
             )
 
         # Basic structural validation - ensure we have at least one field

libs/labelbox/src/labelbox/schema/workflow/workflow.py

@@ -390,16 +390,36 @@ def validate(cls, workflow: "ProjectWorkflow") -> "ProjectWorkflow":
     def update_config(self, reposition: bool = True) -> "ProjectWorkflow":
         """Update the workflow configuration on the server.
 
+        This method automatically validates the workflow before updating to ensure
+        data integrity and prevent invalid configurations from being saved.
+
         Args:
             reposition: Whether to automatically reposition nodes before update
 
         Returns:
             ProjectWorkflow: Updated workflow instance
 
         Raises:
-            ValueError: If the update operation fails
+            ValueError: If validation errors are found or the update operation fails
         """
         try:
+            # Always validate workflow before updating (mandatory for data safety)
+            validation_result = self.check_validity()
+            validation_errors = validation_result.get("errors", [])
+
+            if validation_errors:
+                # Format validation errors for clear user feedback
+                formatted_errors = self.format_validation_errors(
+                    validation_result
+                )
+                logger.error(f"Workflow validation failed: {formatted_errors}")
+
+                # Raise a clear ValueError with validation details
+                raise ValueError(
+                    f"Cannot update workflow configuration due to validation errors:\n{formatted_errors}\n\n"
+                    f"Please fix these issues before updating."
+                )
+
             if reposition:
                 self.reposition_nodes()
 

libs/labelbox/src/labelbox/schema/workflow/workflow_utils.py

@@ -180,7 +180,7 @@ def validate(cls, workflow: "ProjectWorkflow") -> "ProjectWorkflow":
         errors.extend(connection_errors)
 
         # Store validation results
-        workflow._validation_errors = {"validation": errors}
+        workflow._validation_errors = {"errors": errors}
         return workflow
 
     @staticmethod

libs/labelbox/tests/integration/test_workflow.py

@@ -21,7 +21,8 @@
     WorkflowDefinitionId,
     FilterField,
     # Import filter functions
-    created_by,
+    labeled_by,
+    created_by,  # Still works for backward compatibility
     dataset,
     natural_language,
     labeling_time,
@@ -335,6 +336,72 @@ def test_workflow_update_without_reset(client, test_projects):
     assert review_nodes[0].name == "Updated Review"
 
 
+def test_workflow_validation_in_update_config(client, test_projects):
+    """Test the mandatory validation behavior in update_config."""
+    source_project, _ = test_projects
+
+    # Create an invalid workflow (missing connections)
+    workflow = source_project.get_workflow()
+    workflow.reset_config()
+
+    # Add nodes but don't create proper connections (this will be invalid)
+    initial_labeling = workflow.add_node(type=NodeType.InitialLabeling)
+    initial_rework = workflow.add_node(type=NodeType.InitialRework)
+    # Add a review node with no connections - this should cause validation errors
+    review = workflow.add_node(type=NodeType.Review, name="Unconnected Review")
+
+    # Only connect the initial nodes together, leaving review disconnected
+    workflow.add_edge(initial_labeling, initial_rework)  # This is also invalid
+
+    # Test 1: update_config should validate and fail with invalid workflow
+    with pytest.raises(ValueError) as exc_info:
+        workflow.update_config()
+
+    assert "validation errors" in str(exc_info.value).lower()
+    assert "Cannot update workflow configuration" in str(exc_info.value)
+
+    # Test 2: Multiple calls should consistently fail validation
+    with pytest.raises(ValueError) as exc_info:
+        workflow.update_config()
+
+    assert "validation errors" in str(exc_info.value).lower()
+
+    # Test 3: Validation errors should be consistently reported
+    with pytest.raises(ValueError) as exc_info:
+        workflow.update_config()
+
+    # Verify the error message is clear and helpful
+    error_message = str(exc_info.value)
+    assert "validation errors" in error_message.lower()
+    assert "please fix these issues" in error_message.lower()
+
+    # Test 4: Create a valid workflow and test successful update
+    workflow.reset_config()
+
+    initial_labeling = workflow.add_node(type=NodeType.InitialLabeling)
+    initial_rework = workflow.add_node(type=NodeType.InitialRework)
+    review = workflow.add_node(type=NodeType.Review, name="Connected Review")
+    done = workflow.add_node(type=NodeType.Done, name="Final")
+
+    # Create proper connections
+    workflow.add_edge(initial_labeling, review)
+    workflow.add_edge(initial_rework, review)
+    workflow.add_edge(review, done, NodeOutput.Approved)
+
+    # This should work without errors
+    result = workflow.update_config()
+    assert result is not None
+
+    # Test successful update - should not raise any exceptions
+    try:
+        workflow.update_config()
+        # If we get here, the update was successful
+        assert True
+    except ValueError:
+        # Should not happen with a valid workflow
+        assert False, "Valid workflow should not raise validation errors"
+
+
 def test_workflow_copy(client, test_projects):
     """Test copying a workflow between projects."""
     source_project, target_project = test_projects
@@ -353,7 +420,7 @@ def test_workflow_copy(client, test_projects):
     logic = source_workflow.add_node(
         type=NodeType.Logic,
         name="Source Logic",
-        filters=ProjectWorkflowFilter([created_by(["source-user"])]),
+        filters=ProjectWorkflowFilter([labeled_by(["source-user"])]),
     )
     done = source_workflow.add_node(type=NodeType.Done, name="Source Done")
 
@@ -402,7 +469,7 @@ def test_production_logic_node_with_comprehensive_filters(
         match_filters=MatchFilters.Any,
         filters=ProjectWorkflowFilter(
             [
-                created_by(
+                labeled_by(
                     ["cly7gzohg07zz07v5fqs63zmx", "cl7k7a9x1764808vk6bm1hf8e"]
                 ),
                 metadata([m_condition.contains("tag", ["test"])]),
@@ -491,7 +558,9 @@ def test_filter_operations_with_persistence(client, test_projects):
         name="Filter Test",
         filters=ProjectWorkflowFilter(
             [
-                created_by(["user1", "user2"]),
+                created_by(
+                    ["user1", "user2"]
+                ),  # Still works - backward compatibility
                 sample(30),
                 labeling_time.greater_than(500),
             ]
@@ -518,7 +587,7 @@ def test_filter_operations_with_persistence(client, test_projects):
     ), f"Should start with 3 filters, got {initial_count}"
 
     # Test removing filters with persistence
-    logic_node.remove_filter(FilterField.CreatedBy)
+    logic_node.remove_filter(FilterField.LabeledBy)
     logic_node.remove_filter(FilterField.Sample)
     updated_workflow.update_config(reposition=False)
 
@@ -542,7 +611,7 @@ def test_filter_operations_with_persistence(client, test_projects):
     ), "LabelingTime filter should remain"
     assert (
         "CreatedBy" not in remaining_fields
-    ), "CreatedBy filter should be removed"
+    ), "LabeledBy filter should be removed"
 
     # Test adding filters with persistence
     logic_after_removal.add_filter(dataset(["new-dataset"]))