Merge pull request #18517 from mvdbeek/fix_deleted_ipnut_submission_collection

[24.1] Prevent job submission if input collection element is deleted

lib/galaxy/model/__init__.py

@@ -6499,6 +6499,23 @@ class DatasetCollection(Base, Dictifiable, UsesAnnotations, Serializable):
         q = q.order_by(*order_by_columns)
         return q
 
+    @property
+    def elements_deleted(self):
+        if not hasattr(self, "_elements_deleted"):
+            if session := object_session(self):
+                stmt = self._build_nested_collection_attributes_stmt(
+                    hda_attributes=("deleted",), dataset_attributes=("deleted",)
+                )
+                stmt = stmt.exists().where(or_(HistoryDatasetAssociation.deleted == true(), Dataset.deleted == true()))
+                self._elements_deleted = session.execute(select(stmt)).scalar()
+            else:
+                self._elements_deleted = False
+                for dataset_instance in self.dataset_instances:
+                    if dataset_instance.deleted or dataset_instance.dataset.deleted:
+                        self._elements_deleted = True
+                        break
+        return self._elements_deleted
+
     @property
     def dataset_states_and_extensions_summary(self):
         if not hasattr(self, "_dataset_states_and_extensions_summary"):

lib/galaxy/tools/parameters/basic.py

@@ -2189,19 +2189,37 @@ class DataToolParameter(BaseDataToolParameter):
         dataset_matcher_factory = get_dataset_matcher_factory(trans)
         dataset_matcher = dataset_matcher_factory.dataset_matcher(self, other_values)
         for v in rval:
-            if v:
-                if hasattr(v, "deleted") and v.deleted:
+            if isinstance(v, DatasetCollectionElement):
+                if hda := v.hda:
+                    v = hda
+                elif ldda := v.ldda:
+                    v = ldda
+                elif collection := v.child_collection:
+                    v = collection
+                elif not v.collection and v.collection.populated_optimized:
+                    raise ParameterValueError("the selected collection has not been populated.", self.name)
+                else:
+                    raise ParameterValueError("Collection element in unexpected state", self.name)
+            if isinstance(v, DatasetInstance):
+                if v.deleted:
                     raise ParameterValueError("the previously selected dataset has been deleted.", self.name)
-                elif hasattr(v, "dataset") and v.dataset.state in [Dataset.states.ERROR, Dataset.states.DISCARDED]:
+                elif v.dataset and v.dataset.state in [Dataset.states.ERROR, Dataset.states.DISCARDED]:
                     raise ParameterValueError(
                         "the previously selected dataset has entered an unusable state", self.name
                     )
-                elif hasattr(v, "dataset"):
-                    if isinstance(v, DatasetCollectionElement):
-                        v = v.hda
-                    match = dataset_matcher.hda_match(v)
-                    if match and match.implicit_conversion:
-                        v.implicit_conversion = True  # type:ignore[union-attr]
+                match = dataset_matcher.hda_match(v)
+                if match and match.implicit_conversion:
+                    v.implicit_conversion = True  # type:ignore[union-attr]
+            elif isinstance(v, HistoryDatasetCollectionAssociation):
+                if v.deleted:
+                    raise ParameterValueError("the previously selected dataset collection has been deleted.", self.name)
+                v = v.collection
+            if isinstance(v, DatasetCollection):
+                if v.elements_deleted:
+                    raise ParameterValueError(
+                        "the previously selected dataset collection has elements that are deleted.", self.name
+                    )
+
         if not self.multiple:
             if len(rval) > 1:
                 raise ParameterValueError("more than one dataset supplied to single input dataset parameter", self.name)
@@ -2498,10 +2516,19 @@ class DataCollectionToolParameter(BaseDataToolParameter):
                 rval = session.get(HistoryDatasetCollectionAssociation, int(value[len("hdca:") :]))
             else:
                 rval = session.get(HistoryDatasetCollectionAssociation, int(value))
-        if rval and isinstance(rval, HistoryDatasetCollectionAssociation):
-            if rval.deleted:
-                raise ParameterValueError("the previously selected dataset collection has been deleted", self.name)
-            # TODO: Handle error states, implement error states ...
+        if rval:
+            if isinstance(rval, HistoryDatasetCollectionAssociation):
+                if rval.deleted:
+                    raise ParameterValueError("the previously selected dataset collection has been deleted", self.name)
+                if rval.collection.elements_deleted:
+                    raise ParameterValueError(
+                        "the previously selected dataset collection has elements that are deleted.", self.name
+                    )
+            if isinstance(rval, DatasetCollectionElement):
+                if (child_collection := rval.child_collection) and child_collection.elements_deleted:
+                    raise ParameterValueError(
+                        "the previously selected dataset collection has elements that are deleted.", self.name
+                    )
         return rval
 
     def to_text(self, value):

lib/galaxy_test/api/test_jobs.py

@@ -576,6 +576,23 @@ steps:
         if output_dataset_paths_exist:
             wait_on(paths_deleted, "path deletion")
 
+    def test_submission_on_collection_with_deleted_element(self, history_id):
+        hdca = self.dataset_collection_populator.create_list_of_list_in_history(history_id=history_id, wait=True).json()
+        hda_id = hdca["elements"][0]["object"]["elements"][0]["object"]["id"]
+        self.dataset_populator.delete_dataset(history_id=history_id, content_id=hda_id)
+        response = self.dataset_populator.run_tool_raw(
+            "is_of_type",
+            inputs={
+                "collection": {"batch": True, "values": [{"src": "hdca", "id": hdca["id"], "map_over_type": "list"}]},
+            },
+            history_id=history_id,
+        )
+        assert response.status_code == 400
+        assert (
+            response.json()["err_msg"]
+            == "parameter 'collection': the previously selected dataset collection has elements that are deleted."
+        )
+
     @pytest.mark.require_new_history
     @skip_without_tool("create_2")
     def test_purging_output_cleaned_after_ok_run(self, history_id):