Add operator for labeled_by, dataset, issue_category, annotation

Author: paulnoirel

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

@@ -45,7 +45,6 @@
 from labelbox.schema.workflow.project_filter import (
     ProjectWorkflowFilter,
     labeled_by,
-    created_by,  # Deprecated, use labeled_by instead
     annotation,
     dataset,
     issue_category,
@@ -94,13 +93,13 @@
     "NodeType",
     "ProjectWorkflowGraph",
     "ProjectWorkflowFilter",
-    # Filter construction functions
+    # Filter field objects
     "labeled_by",
-    "created_by",  # Deprecated, use labeled_by instead
     "annotation",
-    "sample",
     "dataset",
     "issue_category",
+    # Filter construction functions
+    "sample",
     "model_prediction",
     "natural_language",
     "labeled_at",

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

@@ -101,7 +101,7 @@ def is_one_of(self, values: List[str]) -> Dict[str, Any]:
         Args:
             values: List of IDs to match
         """
-        return {self._field_name: values, "__operator": "is"}
+        return {self._field_name: values}
 
     def is_not_one_of(self, values: List[str]) -> Dict[str, Any]:
         """Filter for items that are NOT one of the specified values.
@@ -111,6 +111,17 @@ def is_not_one_of(self, values: List[str]) -> Dict[str, Any]:
         """
         return {self._field_name: values, "__operator": "is_not"}
 
+    def has_any_of(self, values: List[str]) -> Dict[str, Any]:
+        """Filter for items that have any of the specified values.
+
+        This is semantically equivalent to is_one_of but provides clearer intent
+        for certain filter types like categories or annotations.
+
+        Args:
+            values: List of IDs to match
+        """
+        return {self._field_name: values}
+
 
 class RangeField:
     """Field class for range-based filters like consensus_average."""
@@ -171,30 +182,12 @@ def __call__(
 batch = ListField("Batch")
 consensus_average = RangeField("ConsensusAverage")
 feature_consensus_average = FeatureRangeField("FeatureConsensusAverage")
-# Note: dataset is a function, not a field object
-
-
-# Function versions for filter functions
-def dataset(
-    dataset_ids: List[str], label: Optional[str] = None
-) -> Dict[str, Any]:
-    """Filter by dataset IDs.
 
-    Args:
-        dataset_ids: List of dataset IDs to filter by
-        label: Optional custom label to display in the UI instead of the default "DS-0" format
-
-    Returns:
-        Dict representing the filter rule
-
-    Examples:
-        dataset(["dataset-123", "dataset-456"])
-        dataset(["dataset-123"], label="My Custom Dataset")
-    """
-    result: Dict[str, Any] = {"Dataset": dataset_ids}
-    if label is not None:
-        result["__label"] = label
-    return result
+# List-based filter field instances
+labeled_by = ListField("CreatedBy")  # Maps to backend CreatedBy field
+dataset = ListField("Dataset")
+issue_category = ListField("IssueCategory")
+annotation = ListField("Annotation")
 
 
 class MetadataCondition:
@@ -302,68 +295,6 @@ def metadata(
     return result
 
 
-def labeled_by(
-    user_ids: List[str], label: Optional[str] = None
-) -> Dict[str, Any]:
-    """Filter by users who labeled the data.
-
-    Args:
-        user_ids: List of user IDs
-        label: Optional custom label to display in the UI
-
-    Returns:
-        Dict representing the filter rule
-    """
-    result: Dict[str, Any] = {"CreatedBy": user_ids}
-    if label is not None:
-        result["__label"] = label
-    return result
-
-
-def created_by(
-    user_ids: List[str], label: Optional[str] = None
-) -> Dict[str, Any]:
-    """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
-        label: Optional custom label to display in the UI
-
-    Returns:
-        Dict representing the filter rule
-    """
-    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(
-    schema_node_ids: List[str], label: Optional[str] = None
-) -> Dict[str, Any]:
-    """Filter by annotation schema node IDs.
-
-    Args:
-        schema_node_ids: List of annotation schema node IDs
-        label: Optional custom label to display in the UI
-
-    Returns:
-        Dict representing the filter rule
-    """
-    result: Dict[str, Any] = {"Annotation": schema_node_ids}
-    if label is not None:
-        result["__label"] = label
-    return result
-
-
 def sample(percentage: int, label: Optional[str] = None) -> Dict[str, Any]:
     """Filter by random sample percentage.
 
@@ -391,24 +322,6 @@ def sample(percentage: int, label: Optional[str] = None) -> Dict[str, Any]:
     return result
 
 
-def issue_category(
-    category_ids: List[str], label: Optional[str] = None
-) -> Dict[str, Any]:
-    """Filter by issue category IDs.
-
-    Args:
-        category_ids: List of issue category IDs
-        label: Optional custom label to display in the UI
-
-    Returns:
-        Dict representing the filter rule
-    """
-    result: Dict[str, Any] = {"IssueCategory": category_ids}
-    if label is not None:
-        result["__label"] = label
-    return result
-
-
 def model_prediction(
     conditions: List[Dict[str, Any]], label: Optional[str] = None
 ) -> Dict[str, Any]:
@@ -576,14 +489,17 @@ def convert_to_api_format(filter_rule: Dict[str, Any]) -> Dict[str, Any]:
 
 class ProjectWorkflowFilter(BaseModel):
     """
-    Project workflow filter collection that enforces filter function syntax.
+    Project workflow filter collection that enforces filter syntax.
 
-    Only accepts filters created using filter functions in this module.
+    Only accepts filters created using filter field objects and functions in this module.
     This ensures type safety, IDE support, and eliminates manual string construction errors.
 
     Example Usage:
         filters = ProjectWorkflowFilter([
-            labeled_by(["user-123"]),
+            labeled_by.is_one_of(["user-123"]),
+            dataset.is_one_of(["dataset-456"]),
+            issue_category.has_any_of(["cat1", "cat2"]),
+            annotation.has_any_of(["bbox", "segmentation"]),
             sample(20),
             labeled_at.between("2024-01-01", "2024-12-31"),
             metadata([condition.contains("tag", "test")]),
@@ -594,7 +510,7 @@ class ProjectWorkflowFilter(BaseModel):
         logic.set_filters(filters)
 
         # Or add individual filters
-        logic.add_filter(labeled_by(["user-123"]))
+        logic.add_filter(labeled_by.is_one_of(["user-123"]))
     """
 
     rules: List[Dict[str, Any]] = Field(default_factory=lambda: [])

libs/labelbox/tests/integration/test_workflow.py

@@ -22,7 +22,6 @@
     FilterField,
     # Import filter functions
     labeled_by,
-    created_by,  # Still works for backward compatibility
     dataset,
     natural_language,
     labeling_time,
@@ -420,7 +419,7 @@ def test_workflow_copy(client, test_projects):
     logic = source_workflow.add_node(
         type=NodeType.Logic,
         name="Source Logic",
-        filters=ProjectWorkflowFilter([labeled_by(["source-user"])]),
+        filters=ProjectWorkflowFilter([labeled_by.is_one_of(["source-user"])]),
     )
     done = source_workflow.add_node(type=NodeType.Done, name="Source Done")
 
@@ -469,7 +468,7 @@ def test_production_logic_node_with_comprehensive_filters(
         match_filters=MatchFilters.Any,
         filters=ProjectWorkflowFilter(
             [
-                labeled_by(
+                labeled_by.is_one_of(
                     ["cly7gzohg07zz07v5fqs63zmx", "cl7k7a9x1764808vk6bm1hf8e"]
                 ),
                 metadata([m_condition.contains("tag", ["test"])]),
@@ -480,8 +479,8 @@ def test_production_logic_node_with_comprehensive_filters(
                 ),
                 labeling_time.greater_than(1000),
                 review_time.less_than_or_equal(100),
-                dataset(["cm37vyets000z072314wxgt0l"]),
-                annotation(["cm37w0e0500lf0709ba7c42m9"]),
+                dataset.is_one_of(["cm37vyets000z072314wxgt0l"]),
+                annotation.has_any_of(["cm37w0e0500lf0709ba7c42m9"]),
                 consensus_average(0.17, 0.61),
                 model_prediction(
                     [
@@ -558,9 +557,7 @@ def test_filter_operations_with_persistence(client, test_projects):
         name="Filter Test",
         filters=ProjectWorkflowFilter(
             [
-                created_by(
-                    ["user1", "user2"]
-                ),  # Still works - backward compatibility
+                labeled_by.is_one_of(["user1", "user2"]),  # New syntax
                 sample(30),
                 labeling_time.greater_than(500),
             ]
@@ -614,7 +611,7 @@ def test_filter_operations_with_persistence(client, test_projects):
     ), "LabeledBy filter should be removed"
 
     # Test adding filters with persistence
-    logic_after_removal.add_filter(dataset(["new-dataset"]))
+    logic_after_removal.add_filter(dataset.is_one_of(["new-dataset"]))
     logic_after_removal.add_filter(
         metadata([m_condition.starts_with("priority", "high")])
     )