Shared models mongofy

bia-shared-datamodels/src/bia_shared_datamodels/bia_data_model.py

@@ -1,23 +1,55 @@
 from __future__ import annotations
 
-from . import semantic_models
+from . import semantic_models, exceptions
 from pydantic import BaseModel, Field, ConfigDict
 from typing import List, Optional
 from uuid import UUID
 from enum import Enum
 
 
+class ModelMetadata(BaseModel):
+    type_name: str = Field()
+    version: int = Field()
 
 
 class DocumentMixin(BaseModel):
-
-    # Throw error if you try to validate/create model from a dictionary with keys that aren't a field in the model
-    model_config = ConfigDict(extra="forbid")
-
     uuid: UUID = Field(
         description="""Unique ID (across the BIA database) used to refer to and identify a document."""
     )
 
+    version: int = Field(
+        description="""Document version. This can't be optional to make sure we never persist objects without it"""
+    )
+    model: Optional[ModelMetadata] = Field(
+        description="""Model type and version. Used to map arbitrary objects to a known (possibly previously-used) type.
+        Optional because for some usecases (e.g. api) we want to accept objects without it because we have the info we need to set it."""
+    )
+
+    # Throw error if you try to validate/create model from a dictionary with keys that aren't a field in the model
+    model_config = ConfigDict(extra="forbid")
+
+    def __init__(self, *args, **data):
+        model_version_spec = self.model_config.get("model_version_latest")
+        if model_version_spec is None:
+            raise exceptions.ModelDefinitionInvalid(
+                f"Class {self.__class__.__name__} missing 'model_version_latest' in its model_config"
+            )
+
+        model_metadata_expected = ModelMetadata(
+            type_name=self.__class__.__name__,
+            version=model_version_spec,
+        )
+        model_metadata_existing = data.get("model", None)
+        if model_metadata_existing:
+            if model_metadata_existing != model_metadata_expected:
+                raise exceptions.UnexpectedDocumentType(
+                    f"Document {str(data.get('uuid'))} has model metadata {model_metadata_existing}, expected : {model_metadata_expected}"
+                )
+        else:
+            data["model"] = model_metadata_expected.model_dump()
+
+        super().__init__(*args, **data)
+
 
 class UserIdentifiedObject(BaseModel):
     title_id: str = Field(
@@ -31,13 +63,17 @@ class Study(
 ):
     author: List[semantic_models.Contributor] = Field(min_length=1)
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class FileReference(
     semantic_models.FileReference,
     DocumentMixin,
 ):
     submission_dataset_uuid: UUID = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class ImageRepresentation(
     semantic_models.ImageRepresentation,
@@ -47,6 +83,8 @@ class ImageRepresentation(
     original_file_reference_uuid: Optional[List[UUID]] = Field()
     representation_of_uuid: UUID = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class ExperimentalImagingDataset(
     semantic_models.ExperimentalImagingDataset,
@@ -55,12 +93,16 @@ class ExperimentalImagingDataset(
 ):
     submitted_in_study_uuid: UUID = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class Specimen(semantic_models.Specimen, DocumentMixin):
     imaging_preparation_protocol_uuid: List[UUID] = Field(min_length=1)
     sample_of_uuid: List[UUID] = Field(min_length=1)
     growth_protocol_uuid: List[UUID] = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class ExperimentallyCapturedImage(
     semantic_models.ExperimentallyCapturedImage,
@@ -70,36 +112,39 @@ class ExperimentallyCapturedImage(
     submission_dataset_uuid: UUID = Field()
     subject_uuid: UUID = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
+
 class ImageAcquisition(
     semantic_models.ImageAcquisition,
     DocumentMixin,
     UserIdentifiedObject,
 ):
-    pass
+    model_config = ConfigDict(model_version_latest=1)
 
 
 class SpecimenImagingPrepartionProtocol(
     semantic_models.SpecimenImagingPrepartionProtocol,
     DocumentMixin,
     UserIdentifiedObject,
 ):
-    pass
+    model_config = ConfigDict(model_version_latest=1)
 
 
 class SpecimenGrowthProtocol(
     semantic_models.SpecimenGrowthProtocol,
     DocumentMixin,
     UserIdentifiedObject,
 ):
-    pass
+    model_config = ConfigDict(model_version_latest=1)
 
 
 class BioSample(
     semantic_models.BioSample,
     DocumentMixin,
     UserIdentifiedObject,
 ):
-    pass
+    model_config = ConfigDict(model_version_latest=1)
 
 
 class ImageAnnotationDataset(
@@ -109,6 +154,8 @@ class ImageAnnotationDataset(
 ):
     submitted_in_study_uuid: UUID = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class AnnotationFileReference(
     semantic_models.AnnotationFileReference,
@@ -118,6 +165,8 @@ class AnnotationFileReference(
     source_image_uuid: List[UUID] = Field()
     creation_process_uuid: List[UUID] = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class DerivedImage(
     semantic_models.DerivedImage,
@@ -127,11 +176,12 @@ class DerivedImage(
     submission_dataset_uuid: UUID = Field()
     creation_process_uuid: List[UUID] = Field()
 
+    model_config = ConfigDict(model_version_latest=1)
+
 
 class AnnotationMethod(
     semantic_models.AnnotationMethod,
     DocumentMixin,
     UserIdentifiedObject,
 ):
-    pass
-
+    model_config = ConfigDict(model_version_latest=1)

bia-shared-datamodels/src/bia_shared_datamodels/exceptions.py

@@ -0,0 +1,5 @@
+class UnexpectedDocumentType(Exception):
+    pass
+
+class ModelDefinitionInvalid(Exception):
+    pass