Merge pull request galaxyproject#17628 from mvdbeek/add_basic_model_i…

…mport_validation

[24.0] Add basic model import attribute validation

lib/galaxy/model/store/__init__.py

@@ -34,6 +34,10 @@
 
 from bdbag import bdbag_api as bdb
 from boltons.iterutils import remap
+from pydantic import (
+    BaseModel,
+    ConfigDict,
+)
 from rocrate.model.computationalworkflow import (
     ComputationalWorkflow,
     WorkflowDescription,
@@ -96,7 +100,10 @@
     get_contributors,
     write_to_file,
 )
-from galaxy.schema.schema import ModelStoreFormat
+from galaxy.schema.schema import (
+    DatasetStateField,
+    ModelStoreFormat,
+)
 from galaxy.security.idencoding import IdEncodingHelper
 from galaxy.util import (
     FILENAME_VALID_CHARS,
@@ -174,6 +181,20 @@ class ImportDiscardedDataType(Enum):
     FORCE = "force"
 
 
+class DatasetAttributeImportModel(BaseModel):
+    state: Optional[DatasetStateField] = None
+    deleted: Optional[bool] = None
+    purged: Optional[bool] = None
+    external_filename: Optional[str] = None
+    _extra_files_path: Optional[str] = None
+    file_size: Optional[int] = None
+    object_store_id: Optional[str] = None
+    total_size: Optional[int] = None
+    created_from_basename: Optional[str] = None
+    uuid: Optional[str] = None
+    model_config = ConfigDict(extra="ignore")
+
+
 DEFAULT_DISCARDED_DATA_TYPE = ImportDiscardedDataType.FORBID
 
 
@@ -444,22 +465,11 @@ def _import_datasets(
         def handle_dataset_object_edit(dataset_instance, dataset_attrs):
             if "dataset" in dataset_attrs:
                 assert self.import_options.allow_dataset_object_edit
-                dataset_attributes = [
-                    "state",
-                    "deleted",
-                    "purged",
-                    "external_filename",
-                    "_extra_files_path",
-                    "file_size",
-                    "object_store_id",
-                    "total_size",
-                    "created_from_basename",
-                    "uuid",
-                ]
-
-                for attribute in dataset_attributes:
-                    if attribute in dataset_attrs["dataset"]:
-                        setattr(dataset_instance.dataset, attribute, dataset_attrs["dataset"][attribute])
+                dataset_attributes = DatasetAttributeImportModel(**dataset_attrs["dataset"]).model_dump(
+                    exclude_unset=True,
+                )
+                for attribute, value in dataset_attributes.items():
+                    setattr(dataset_instance.dataset, attribute, value)
                 self._attach_dataset_hashes(dataset_attrs["dataset"], dataset_instance)
                 self._attach_dataset_sources(dataset_attrs["dataset"], dataset_instance)
                 if "id" in dataset_attrs["dataset"] and self.import_options.allow_edit:

lib/galaxy/schema/schema.py

@@ -20,6 +20,7 @@
     AnyHttpUrl,
     AnyUrl,
     BaseModel,
+    BeforeValidator,
     ConfigDict,
     Field,
     Json,
@@ -151,11 +152,11 @@ class DatasetCollectionPopulatedState(str, Enum):
 JobId = Annotated[EncodedDatabaseIdField, Field(..., title="Job ID")]
 
 
-DatasetStateField: DatasetState = Field(
-    ...,
-    title="State",
-    description="The current state of this dataset.",
-)
+DatasetStateField = Annotated[
+    DatasetState,
+    BeforeValidator(lambda v: "discarded" if v == "deleted" else v),
+    Field(..., title="State", description="The current state of this dataset."),
+]
 
 CreateTimeField = Field(
     title="Create Time",
@@ -661,7 +662,7 @@ class HDASummary(HDACommon):
         title="Dataset ID",
         description="The encoded ID of the dataset associated with this item.",
     )
-    state: DatasetState = DatasetStateField
+    state: DatasetStateField
     extension: Optional[str] = Field(
         ...,
         title="Extension",
@@ -679,7 +680,7 @@ class HDAInaccessible(HDACommon):
     """History Dataset Association information when the user can not access it."""
 
     accessible: bool = AccessibleField
-    state: DatasetState = DatasetStateField
+    state: DatasetStateField
 
 
 HdaLddaField = Field(
@@ -872,7 +873,7 @@ class HDAObject(Model, WithModelClass):
     # If so at least merge models
     id: HistoryDatasetAssociationId
     model_class: HDA_MODEL_CLASS = ModelClassField(HDA_MODEL_CLASS)
-    state: DatasetState = DatasetStateField
+    state: DatasetStateField
     hda_ldda: DatasetSourceType = HdaLddaField
     history_id: HistoryID
     tags: List[str]
@@ -3080,7 +3081,7 @@ class FileLibraryFolderItem(LibraryFolderItemBase):
     date_uploaded: datetime
     is_unrestricted: bool
     is_private: bool
-    state: DatasetState = DatasetStateField
+    state: DatasetStateField
     file_size: str
     raw_size: int
     ldda_id: EncodedDatabaseIdField
@@ -3650,7 +3651,7 @@ class DatasetSummary(Model):
     id: EncodedDatabaseIdField
     create_time: Optional[datetime] = CreateTimeField
     update_time: Optional[datetime] = UpdateTimeField
-    state: DatasetState = DatasetStateField
+    state: DatasetStateField
     deleted: bool
     purged: bool
     purgable: bool

test/unit/schema/test_schema.py

@@ -1,5 +1,8 @@
 from uuid import uuid4
 
+from pydantic import BaseModel
+
+from galaxy.schema.schema import DatasetStateField
 from galaxy.schema.tasks import (
     GenerateInvocationDownload,
     RequestUser,
@@ -22,3 +25,12 @@ def test_task_schema():
     assert rehydrated_download.invocation_id == TEST_INVOCATION_ID
     assert rehydrated_download.user.user_id == TEST_USER_ID
     assert rehydrated_download.galaxy_url == TEST_GALAXY_URL
+
+
+class StateModel(BaseModel):
+    state: DatasetStateField
+
+
+def test_dataset_state_coercion():
+    assert StateModel(state="ok").state == "ok"
+    assert StateModel(state="deleted").state == "discarded"