feedback

Author: Gabefire

libs/labelbox/src/labelbox/data/annotation_types/data/conversation.py

@@ -3,5 +3,5 @@
 from .base_data import BaseData
 
 
-class ConversationData(BaseData):
-    pass
+class ConversationData(BaseData, _NoCoercionMixin):
+    class_name: Literal["DicomData"] = "ConversationData"

libs/labelbox/src/labelbox/data/serialization/ndjson/base.py

@@ -3,19 +3,21 @@
 
 from labelbox.utils import _CamelCaseMixin, is_exactly_one_set
 from ...annotation_types.types import Cuid
-from pydantic import field_validator, model_validator, model_serializer, ConfigDict, BaseModel, Field
-from uuid import UUID, uuid4
+from pydantic import model_validator, ConfigDict, BaseModel, Field
+from uuid import uuid4
+import threading
 
 subclass_registry = {}
 
-class SubclassRegistryBase(BaseModel):
+class _SubclassRegistryBase(BaseModel):
 
     model_config = ConfigDict(extra="allow")
 
     def __init_subclass__(cls, **kwargs):
         super().__init_subclass__(**kwargs)
         if cls.__name__ != "NDAnnotation":
-            subclass_registry[cls.__name__] = cls  
+            with threading.Lock():
+                subclass_registry[cls.__name__] = cls  
 
 class DataRow(_CamelCaseMixin):
     id: Optional[str] = None

libs/labelbox/src/labelbox/data/serialization/ndjson/classification.py

@@ -11,7 +11,7 @@
 from ...annotation_types.data import TextData, VideoData, ImageData
 from pydantic import model_validator, Field, BaseModel, ConfigDict, model_serializer
 from pydantic.alias_generators import to_camel
-from .base import SubclassRegistryBase
+from .base import _SubclassRegistryBase
 
 
 class NDAnswer(ConfidenceMixin, CustomMetricsMixin):
@@ -170,7 +170,7 @@ def from_common(cls, prompt_text: PromptText, name: str,
 # ====== End of subclasses
 
 
-class NDText(NDAnnotation, NDTextSubclass, SubclassRegistryBase):
+class NDText(NDAnnotation, NDTextSubclass, _SubclassRegistryBase):
 
     @classmethod
     def from_common(cls,
@@ -194,7 +194,7 @@ def from_common(cls,
         )
 
 
-class NDChecklist(NDAnnotation, NDChecklistSubclass, VideoSupported, SubclassRegistryBase):
+class NDChecklist(NDAnnotation, NDChecklistSubclass, VideoSupported, _SubclassRegistryBase):
 
     @model_serializer(mode="wrap")
     def serialize_model(self, handler):
@@ -237,7 +237,7 @@ def from_common(
                    confidence=confidence)
 
 
-class NDRadio(NDAnnotation, NDRadioSubclass, VideoSupported, SubclassRegistryBase):
+class NDRadio(NDAnnotation, NDRadioSubclass, VideoSupported, _SubclassRegistryBase):
 
     @classmethod
     def from_common(
@@ -275,7 +275,7 @@ def serialize_model(self, handler):
         return res
 
 
-class NDPromptText(NDAnnotation, NDPromptTextSubclass, SubclassRegistryBase):
+class NDPromptText(NDAnnotation, NDPromptTextSubclass, _SubclassRegistryBase):
 
     @classmethod
     def from_common(

libs/labelbox/src/labelbox/data/serialization/ndjson/converter.py

@@ -34,8 +34,7 @@ def deserialize(json_data: Iterable[Dict[str, Any]]) -> LabelGenerator:
         Returns:
             LabelGenerator containing the ndjson data.
         """
-        data = copy.copy(json_data)
-        data = NDLabel(**{"annotations": data})
+        data = NDLabel(**{"annotations": copy.copy(json_data)})
         res = data.to_common()
         return res
 

libs/labelbox/src/labelbox/data/serialization/ndjson/label.py

@@ -2,9 +2,6 @@
 from operator import itemgetter
 from typing import Dict, Generator, List, Tuple, Union
 from collections import defaultdict
-from typing_extensions import Unpack
-import warnings
-
 
 from ...annotation_types.annotation import ClassificationAnnotation, ObjectAnnotation
 from ...annotation_types.relationship import RelationshipAnnotation
@@ -23,10 +20,9 @@
 from .objects import NDObject, NDObjectType, NDSegments, NDDicomSegments, NDVideoMasks, NDDicomMasks
 from .relationship import NDRelationship
 from .base import DataRow
-from pydantic import BaseModel, ConfigDict, model_serializer, ValidationError
-from .base import subclass_registry, SubclassRegistryBase, NDAnnotation
+from pydantic import BaseModel, ValidationError
+from .base import subclass_registry, _SubclassRegistryBase
 from pydantic_core import PydanticUndefined
-from pydantic.alias_generators import to_camel
 from contextlib import suppress
 
 AnnotationType = Union[NDObjectType, NDClassificationType, NDPromptClassificationType,
@@ -36,48 +32,51 @@
 
 
 class NDLabel(BaseModel):
-    annotations: List[SubclassRegistryBase]
+    annotations: List[_SubclassRegistryBase]
 
     def __init__(self, **kwargs):
         # NOTE: Deserialization of subclasses in pydantic is difficult, see here https://blog.devgenius.io/deserialize-child-classes-with-pydantic-that-gonna-work-784230e1cf83
         # Below implements the subclass registry as mentioned in the article. The python dicts we pass in can be missing certain fields
-        # we essentially have to infer the type against all sub classes that have the SubclasssRegistryBase inheritance. 
+        # we essentially have to infer the type against all sub classes that have the _SubclasssRegistryBase inheritance. 
         # It works by checking if the keys of our annotations we are missing in matches any required subclass.
         # More keys are prioritized over less keys (closer match). This is used when importing json to our base models not a lot of customer workflows
         # depend on this method but this works for all our existing tests with the bonus of added validation. (no subclass found it throws an error)
-        # Previous strategies hacked but dont work for pydantic V2
-        for index in range(len(kwargs["annotations"])):
-            annotation = kwargs["annotations"][index]
+
+        for index, annotation in enumerate(kwargs["annotations"]):
             if isinstance(annotation, dict):
                 item_annotation_keys = annotation.keys()
                 key_subclass_combos = defaultdict(list)
-                for name, subclass in subclass_registry.items():
-                    subclass: BaseModel = subclass
+                for subclass in subclass_registry.values():
 
                     # Get all required keys from subclass
                     annotation_keys = []
                     for k, field in subclass.model_fields.items():
-                        # must account for alias
-                        if hasattr(field, "validation_alias") and field.validation_alias == "answers" and "answers" in item_annotation_keys:
-                            annotation_keys.append("answers")
-                        elif field.default == PydanticUndefined and k != "uuid":
-                            annotation_keys.append(to_camel(k))
+                        if field.default == PydanticUndefined and k != "uuid":
+                            if hasattr(field, "alias") and field.alias in item_annotation_keys:
+                                annotation_keys.append(field.alias)
+                            else:
+                                annotation_keys.append(k)
+                        
                     key_subclass_combos[subclass].extend(annotation_keys)
-
-                # Sort by subclass that has the most keys
+                    
+                # Sort by subclass that has the most keys i.e. the one with the most keys that matches is most likely our subclass
                 key_subclass_combos = dict(sorted(key_subclass_combos.items(), key = lambda x : len(x[1]), reverse=True))
-                
-                # Choose the key that our dict we are passing in has all the keys to match
+
                 for subclass, key_subclass_combo in key_subclass_combos.items():
+                    # Choose the keys from our dict we supplied that matches the required keys of a subclass
                     check_required_keys = all(key in list(item_annotation_keys) for key in key_subclass_combo)
                     if check_required_keys:
-                        # Keep trying subclasses until we find one that has valid values
+                        # Keep trying subclasses until we find one that has valid values (does not throw an validation error)
                         with suppress(ValidationError):
                             annotation = subclass(**annotation)
                             break
-                kwargs["annotations"][index] = annotation
+                if isinstance(annotation, dict):
+                    raise ValueError(f"Could not find subclass for fields: {item_annotation_keys}")
+                
+            kwargs["annotations"][index] = annotation
         super().__init__(**kwargs)
 
+
     class _Relationship(BaseModel):
         """This object holds information about the relationship"""
         ndjson: NDRelationship

libs/labelbox/src/labelbox/data/serialization/ndjson/metric.py

@@ -9,7 +9,7 @@
     ConfusionMatrixAggregation, ConfusionMatrixMetric,
     ConfusionMatrixMetricValue, ConfusionMatrixMetricConfidenceValue)
 from pydantic import ConfigDict, model_serializer
-from .base import SubclassRegistryBase
+from .base import _SubclassRegistryBase
 
 
 class BaseNDMetric(NDJsonBase):
@@ -27,7 +27,7 @@ def serialize_model(self, handler):
         return res
 
 
-class NDConfusionMatrixMetric(BaseNDMetric, SubclassRegistryBase):
+class NDConfusionMatrixMetric(BaseNDMetric, _SubclassRegistryBase):
     metric_value: Union[ConfusionMatrixMetricValue,
                         ConfusionMatrixMetricConfidenceValue]
     metric_name: str
@@ -54,7 +54,7 @@ def from_common(
                    data_row=DataRow(id=data.uid, global_key=data.global_key))
 
 
-class NDScalarMetric(BaseNDMetric, SubclassRegistryBase):
+class NDScalarMetric(BaseNDMetric, _SubclassRegistryBase):
     metric_value: Union[ScalarMetricValue, ScalarMetricConfidenceValue]
     metric_name: Optional[str] = None
     aggregation: Optional[ScalarMetricAggregation] = ScalarMetricAggregation.ARITHMETIC_MEAN

libs/labelbox/src/labelbox/data/serialization/ndjson/objects.py

@@ -19,7 +19,7 @@
 from ...annotation_types.annotation import ClassificationAnnotation, ObjectAnnotation
 from ...annotation_types.video import VideoMaskAnnotation, DICOMMaskAnnotation, MaskFrame, MaskInstance
 from .classification import NDClassification, NDSubclassification, NDSubclassificationType
-from .base import DataRow, NDAnnotation, NDJsonBase, SubclassRegistryBase
+from .base import DataRow, NDAnnotation, NDJsonBase, _SubclassRegistryBase
 from pydantic import BaseModel
 
 
@@ -48,7 +48,7 @@ class Bbox(BaseModel):
     width: float
 
 
-class NDPoint(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDPoint(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     point: _Point
 
     def to_common(self) -> Point:
@@ -79,7 +79,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDFramePoint(VideoSupported, SubclassRegistryBase):
+class NDFramePoint(VideoSupported, _SubclassRegistryBase):
     point: _Point
     classifications: List[NDSubclassificationType] = []
 
@@ -109,7 +109,7 @@ def from_common(
                    classifications=classifications)
 
 
-class NDLine(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDLine(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     line: List[_Point]
 
     def to_common(self) -> Line:
@@ -140,7 +140,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDFrameLine(VideoSupported, SubclassRegistryBase):
+class NDFrameLine(VideoSupported, _SubclassRegistryBase):
     line: List[_Point]
     classifications: List[NDSubclassificationType] = []
 
@@ -173,7 +173,7 @@ def from_common(
                    classifications=classifications)
 
 
-class NDDicomLine(NDFrameLine, SubclassRegistryBase):
+class NDDicomLine(NDFrameLine, _SubclassRegistryBase):
 
     def to_common(self, name: str, feature_schema_id: Cuid, segment_index: int,
                   group_key: str) -> DICOMObjectAnnotation:
@@ -187,7 +187,7 @@ def to_common(self, name: str, feature_schema_id: Cuid, segment_index: int,
             group_key=group_key)
 
 
-class NDPolygon(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDPolygon(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     polygon: List[_Point]
 
     def to_common(self) -> Polygon:
@@ -218,7 +218,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDRectangle(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDRectangle(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     bbox: Bbox
 
     def to_common(self) -> Rectangle:
@@ -254,7 +254,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDDocumentRectangle(NDRectangle, SubclassRegistryBase):
+class NDDocumentRectangle(NDRectangle, _SubclassRegistryBase):
     page: int
     unit: str
 
@@ -293,7 +293,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDFrameRectangle(VideoSupported, SubclassRegistryBase):
+class NDFrameRectangle(VideoSupported, _SubclassRegistryBase):
     bbox: Bbox
     classifications: List[NDSubclassificationType] = []
 
@@ -398,7 +398,7 @@ def to_common(self, name: str, feature_schema_id: Cuid, uuid: str,
         ]
 
 
-class NDSegments(NDBaseObject, SubclassRegistryBase):
+class NDSegments(NDBaseObject, _SubclassRegistryBase):
     segments: List[NDSegment]
 
     def to_common(self, name: str, feature_schema_id: Cuid):
@@ -425,7 +425,7 @@ def from_common(cls, segments: List[VideoObjectAnnotation], data: VideoData,
                    uuid=extra.get('uuid'))
 
 
-class NDDicomSegments(NDBaseObject, DicomSupported, SubclassRegistryBase):
+class NDDicomSegments(NDBaseObject, DicomSupported, _SubclassRegistryBase):
     segments: List[NDDicomSegment]
 
     def to_common(self, name: str, feature_schema_id: Cuid):
@@ -463,7 +463,7 @@ class _PNGMask(BaseModel):
     png: str
 
 
-class NDMask(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDMask(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     mask: Union[_URIMask, _PNGMask]
 
     def to_common(self) -> Mask:
@@ -517,7 +517,7 @@ class NDVideoMasksFramesInstances(BaseModel):
     instances: List[MaskInstance]
 
 
-class NDVideoMasks(NDJsonBase, ConfidenceMixin, CustomMetricsNotSupportedMixin, SubclassRegistryBase):
+class NDVideoMasks(NDJsonBase, ConfidenceMixin, CustomMetricsNotSupportedMixin, _SubclassRegistryBase):
     masks: NDVideoMasksFramesInstances
 
     def to_common(self) -> VideoMaskAnnotation:
@@ -545,7 +545,7 @@ def from_common(cls, annotation, data):
         )
 
 
-class NDDicomMasks(NDVideoMasks, DicomSupported, SubclassRegistryBase):
+class NDDicomMasks(NDVideoMasks, DicomSupported, _SubclassRegistryBase):
 
     def to_common(self) -> DICOMMaskAnnotation:
         return DICOMMaskAnnotation(
@@ -569,7 +569,7 @@ class Location(BaseModel):
     end: int
 
 
-class NDTextEntity(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDTextEntity(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     location: Location
 
     def to_common(self) -> TextEntity:
@@ -601,7 +601,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDDocumentEntity(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, SubclassRegistryBase):
+class NDDocumentEntity(NDBaseObject, ConfidenceMixin, CustomMetricsMixin, _SubclassRegistryBase):
     name: str
     text_selections: List[DocumentTextSelection]
 
@@ -633,7 +633,7 @@ def from_common(
                    custom_metrics=custom_metrics)
 
 
-class NDConversationEntity(NDTextEntity, SubclassRegistryBase):
+class NDConversationEntity(NDTextEntity, _SubclassRegistryBase):
     message_id: str
 
     def to_common(self) -> ConversationEntity:

libs/labelbox/src/labelbox/data/serialization/ndjson/relationship.py

@@ -5,7 +5,7 @@
 from ...annotation_types.relationship import RelationshipAnnotation
 from ...annotation_types.relationship import Relationship
 from .objects import NDObjectType
-from .base import DataRow, SubclassRegistryBase
+from .base import DataRow, _SubclassRegistryBase
 
 SUPPORTED_ANNOTATIONS = NDObjectType
 
@@ -16,7 +16,7 @@ class _Relationship(BaseModel):
     type: str
 
 
-class NDRelationship(NDAnnotation, SubclassRegistryBase):
+class NDRelationship(NDAnnotation, _SubclassRegistryBase):
     relationship: _Relationship
 
     @staticmethod