fix: robust synthetic property handling for experimentst

- Only generate synthetic fields for experiments (e.g., publication identifiers, score set URNs) when ORM attributes are present, avoiding dict-based synthesis.
- Validators now check for ORM attribute presence before transformation, ensuring correct behavior for both ORM and API/dict contexts.
- Updated tests to expect Pydantic validation errors when required synthetic fields are missing.

Author: bencap

src/mavedb/view_models/experiment.py

@@ -1,15 +1,15 @@
 from datetime import date
 from typing import Any, Collection, Optional, Sequence
 
-from pydantic import field_validator, model_validator, ValidationInfo
+from pydantic import ValidationInfo, field_validator, model_validator
 
+from mavedb.lib.validation import urn_re
 from mavedb.lib.validation.exceptions import ValidationError
 from mavedb.lib.validation.transform import (
     transform_experiment_set_to_urn,
-    transform_score_set_list_to_urn_list,
     transform_record_publication_identifiers,
+    transform_score_set_list_to_urn_list,
 )
-from mavedb.lib.validation import urn_re
 from mavedb.lib.validation.utilities import is_null
 from mavedb.view_models import record_type_validator, set_record_type
 from mavedb.view_models.base.base import BaseModel
@@ -129,12 +129,11 @@ def publication_identifiers_validator(cls, v: Any, info: ValidationInfo) -> list
         return list(v)  # Re-cast into proper list-like type
 
     # These 'synthetic' fields are generated from other model properties. Transform data from other properties as needed, setting
-    # the appropriate field on the model itself. Then, proceed with Pydantic ingestion once fields are created.
+    # the appropriate field on the model itself. Then, proceed with Pydantic ingestion once fields are created. Only perform these
+    # transformations if the relevant attributes are present on the input data (i.e., when creating from an ORM object).
     @model_validator(mode="before")
     def generate_primary_and_secondary_publications(cls, data: Any):
-        if not hasattr(data, "primary_publication_identifiers") or not hasattr(
-            data, "secondary_publication_identifiers"
-        ):
+        if hasattr(data, "publication_identifier_associations"):
             try:
                 publication_identifiers = transform_record_publication_identifiers(
                     data.publication_identifier_associations
@@ -145,28 +144,30 @@ def generate_primary_and_secondary_publications(cls, data: Any):
                 data.__setattr__(
                     "secondary_publication_identifiers", publication_identifiers["secondary_publication_identifiers"]
                 )
-            except AttributeError as exc:
+            except (KeyError, AttributeError) as exc:
                 raise ValidationError(
-                    f"Unable to create {cls.__name__} without attribute: {exc}."  # type: ignore
+                    f"Unable to coerce publication identifier attributes from ORM for {cls.__name__}: {exc}."  # type: ignore
                 )
         return data
 
     @model_validator(mode="before")
     def generate_score_set_urn_list(cls, data: Any):
-        if not hasattr(data, "score_set_urns"):
+        if hasattr(data, "score_sets"):
             try:
                 data.__setattr__("score_set_urns", transform_score_set_list_to_urn_list(data.score_sets))
-            except AttributeError as exc:
-                raise ValidationError(f"Unable to create {cls.__name__} without attribute: {exc}.")  # type: ignore
+            except (KeyError, AttributeError) as exc:
+                raise ValidationError(f"Unable to coerce associated score set URNs from ORM for {cls.__name__}: {exc}.")  # type: ignore
         return data
 
     @model_validator(mode="before")
     def generate_experiment_set_urn(cls, data: Any):
-        if not hasattr(data, "experiment_set_urn"):
+        if hasattr(data, "experiment_set"):
             try:
                 data.__setattr__("experiment_set_urn", transform_experiment_set_to_urn(data.experiment_set))
-            except AttributeError as exc:
-                raise ValidationError(f"Unable to create {cls.__name__} without attribute: {exc}.")  # type: ignore
+            except (KeyError, AttributeError) as exc:
+                raise ValidationError(
+                    f"Unable to coerce associated experiment set URN from ORM for {cls.__name__}: {exc}."
+                )  # type: ignore
         return data
 
 

tests/view_models/test_experiment.py

@@ -1,17 +1,18 @@
 import pytest
+from pydantic import ValidationError
 
-from mavedb.view_models.experiment import ExperimentCreate, SavedExperiment
+from mavedb.view_models.experiment import Experiment, ExperimentCreate, SavedExperiment
 from mavedb.view_models.publication_identifier import PublicationIdentifier
-
 from tests.helpers.constants import (
-    VALID_EXPERIMENT_URN,
-    VALID_SCORE_SET_URN,
-    VALID_EXPERIMENT_SET_URN,
+    SAVED_BIORXIV_PUBLICATION,
+    SAVED_PUBMED_PUBLICATION,
+    TEST_BIORXIV_IDENTIFIER,
     TEST_MINIMAL_EXPERIMENT,
     TEST_MINIMAL_EXPERIMENT_RESPONSE,
-    SAVED_PUBMED_PUBLICATION,
     TEST_PUBMED_IDENTIFIER,
-    TEST_BIORXIV_IDENTIFIER,
+    VALID_EXPERIMENT_SET_URN,
+    VALID_EXPERIMENT_URN,
+    VALID_SCORE_SET_URN,
 )
 from tests.helpers.util.common import dummy_attributed_object_from_dict
 
@@ -237,8 +238,15 @@ def test_saved_experiment_synthetic_properties():
     )
 
 
-@pytest.mark.parametrize("exclude", ["publication_identifier_associations", "score_sets", "experiment_set"])
-def test_cannot_create_saved_experiment_without_all_attributed_properties(exclude):
+@pytest.mark.parametrize(
+    "exclude,expected_missing_fields",
+    [
+        ("publication_identifier_associations", ["primaryPublicationIdentifiers", "secondaryPublicationIdentifiers"]),
+        ("score_sets", ["scoreSetUrns"]),
+        ("experiment_set", ["experimentSetUrn"]),
+    ],
+)
+def test_cannot_create_saved_experiment_without_all_attributed_properties(exclude, expected_missing_fields):
     experiment = TEST_MINIMAL_EXPERIMENT_RESPONSE.copy()
     experiment["urn"] = VALID_EXPERIMENT_URN
 
@@ -280,11 +288,14 @@ def test_cannot_create_saved_experiment_without_all_attributed_properties(exclud
 
     experiment.pop(exclude)
     experiment_attributed_object = dummy_attributed_object_from_dict(experiment)
-    with pytest.raises(ValueError) as exc_info:
+    with pytest.raises(ValidationError) as exc_info:
         SavedExperiment.model_validate(experiment_attributed_object)
 
-    assert "Unable to create SavedExperiment without attribute" in str(exc_info.value)
-    assert exclude in str(exc_info.value)
+    # Should fail with missing fields coerced from missing attributed properties
+    msg = str(exc_info.value)
+    assert "Field required" in msg
+    for field in expected_missing_fields:
+        assert field in msg
 
 
 def test_can_create_experiment_with_nonetype_experiment_set_urn():
@@ -303,3 +314,20 @@ def test_cant_create_experiment_with_invalid_experiment_set_urn():
         ExperimentCreate(**experiment_test)
 
     assert f"'{experiment_test['experiment_set_urn']}' is not a valid experiment set URN" in str(exc_info.value)
+
+
+def test_can_create_experiment_from_non_orm_context():
+    experiment = TEST_MINIMAL_EXPERIMENT_RESPONSE.copy()
+    experiment["urn"] = VALID_EXPERIMENT_URN
+    experiment["experimentSetUrn"] = VALID_EXPERIMENT_SET_URN
+    experiment["scoreSetUrns"] = [VALID_SCORE_SET_URN]
+    experiment["primaryPublicationIdentifiers"] = [SAVED_PUBMED_PUBLICATION]
+    experiment["secondaryPublicationIdentifiers"] = [SAVED_PUBMED_PUBLICATION, SAVED_BIORXIV_PUBLICATION]
+
+    # Should not require any ORM attributes
+    saved_experiment = Experiment.model_validate(experiment)
+    assert saved_experiment.urn == VALID_EXPERIMENT_URN
+    assert saved_experiment.experiment_set_urn == VALID_EXPERIMENT_SET_URN
+    assert saved_experiment.score_set_urns == [VALID_SCORE_SET_URN]
+    assert len(saved_experiment.primary_publication_identifiers) == 1
+    assert len(saved_experiment.secondary_publication_identifiers) == 2