Merge pull request #17071 from mvdbeek/assert_dataset_collection_element_points_at_something

[23.1] Assert that ``DatasetCollectioElement`` has an associated object

lib/galaxy/model/__init__.py

@@ -4918,6 +4918,20 @@ class HistoryDatasetAssociation(DatasetInstance, HasTags, Dictifiable, UsesAnnot
         self.copied_from_history_dataset_association = copied_from_history_dataset_association
         self.copied_from_library_dataset_dataset_association = copied_from_library_dataset_dataset_association
 
+    def __strict_check_before_flush__(self):
+        if self.extension != "len":
+            # TODO: Custom builds (with .len extension) do not get a history or a HID.
+            # These should get some other type of permanent storage, perhaps UserDatasetAssociation ?
+            # Everything else needs to have a hid and a history
+            if not self.history and not getattr(self, "history_id", None):
+                raise Exception(f"HistoryDatasetAssociation {self} without history detected, this is not valid")
+            elif not self.hid:
+                raise Exception(f"HistoryDatasetAssociation {self} without hid, this is not valid")
+            elif self.dataset.file_size is None and self.dataset.state not in self.dataset.no_data_states:
+                raise Exception(
+                    f"HistoryDatasetAssociation {self} in state {self.dataset.state} with null file size, this is not valid"
+                )
+
     @property
     def user(self):
         if self.history:
@@ -6995,6 +7009,9 @@ class DatasetCollectionElement(Base, Dictifiable, Serializable):
         self.element_index = element_index
         self.element_identifier = element_identifier or str(element_index)
 
+    def __strict_check_before_flush__(self):
+        assert self.element_object, "Dataset Collection Element without child entity detected, this is not valid"
+
     @property
     def element_type(self):
         if self.hda:

lib/galaxy/model/base.py

@@ -3,6 +3,7 @@ Shared model and mapping code between Galaxy and Tool Shed, trying to
 generalize to generic database connections.
 """
 import contextlib
+import logging
 import os
 import threading
 from contextvars import ContextVar
@@ -29,6 +30,8 @@ from galaxy.util.bunch import Bunch
 if TYPE_CHECKING:
     from galaxy.model.store import SessionlessContext
 
+log = logging.getLogger(__name__)
+
 # Create a ContextVar with mutable state, this allows sync tasks in the context
 # of a request (which run within a threadpool) to see changes to the ContextVar
 # state. See https://github.com/tiangolo/fastapi/issues/953#issuecomment-586006249
@@ -146,23 +149,14 @@ def versioned_objects(iter):
 
 def versioned_objects_strict(iter):
     for obj in iter:
+        if hasattr(obj, "__strict_check_before_flush__"):
+            obj.__strict_check_before_flush__()
         if hasattr(obj, "__create_version__"):
-            if obj.extension != "len":
-                # TODO: Custom builds (with .len extension) do not get a history or a HID.
-                # These should get some other type of permanent storage, perhaps UserDatasetAssociation ?
-                # Everything else needs to have a hid and a history
-                if not obj.history and not obj.history_id:
-                    raise Exception(f"HistoryDatasetAssociation {obj} without history detected, this is not valid")
-                elif not obj.hid:
-                    raise Exception(f"HistoryDatasetAssociation {obj} without hid, this is not valid")
-                elif obj.dataset.file_size is None and obj.dataset.state not in obj.dataset.no_data_states:
-                    raise Exception(
-                        f"HistoryDatasetAssociation {obj} in state {obj.dataset.state} with null file size, this is not valid"
-                    )
             yield obj
 
 
 if os.environ.get("GALAXY_TEST_RAISE_EXCEPTION_ON_HISTORYLESS_HDA"):
+    log.debug("Using strict flush checks")
     versioned_objects = versioned_objects_strict  # noqa: F811
 
 

lib/galaxy_test/driver/driver_util.py

@@ -877,6 +877,8 @@ class GalaxyTestDriver(TestDriver):
         """Setup various variables used to launch a Galaxy server."""
         config_object = self._ensure_config_object(config_object)
         self.external_galaxy = os.environ.get("GALAXY_TEST_EXTERNAL", None)
+        if not self.external_galaxy:
+            os.environ["GALAXY_TEST_STRICT_CHECKS"] = "1"
 
         # Allow controlling the log format
         self.log_format = os.environ.get("GALAXY_TEST_LOG_FORMAT")