Delete task parents m2m before tasks in corpus deletion

arkindex/documents/tasks.py

@@ -75,6 +75,8 @@ def corpus_delete(corpus_id: str) -> None:
         ProcessDataset.objects.filter(process__corpus_id=corpus_id),
         DatasetElement.objects.filter(dataset__corpus_id=corpus_id),
         corpus.datasets.all(),
+        # Delete the hidden M2M task parents table
+        Task.parents.through.objects.filter(from_task__process__corpus_id=corpus_id),
         Task.objects.filter(process__corpus_id=corpus_id),
         corpus.elements.all(),
         WorkerRun.objects.filter(process__corpus_id=corpus_id),

arkindex/documents/tests/tasks/test_corpus_delete.py

@@ -2,7 +2,7 @@ from django.db.models.signals import pre_delete
 
 from arkindex.documents.models import Corpus, Element, EntityType, MetaType, Transcription
 from arkindex.documents.tasks import corpus_delete
-from arkindex.ponos.models import Farm
+from arkindex.ponos.models import Farm, State, Task
 from arkindex.process.models import CorpusWorkerVersion, ProcessMode, Repository, WorkerVersion
 from arkindex.project.tests import FixtureTestCase, force_constraints_immediate
 from arkindex.training.models import Dataset
@@ -42,6 +42,20 @@ class TestDeleteCorpus(FixtureTestCase):
         )
         element_process.elements.add(element)
         worker_run = element_process.worker_runs.create(version=cls.worker_version, parents=[])
+        task_1, task_2, task_3 = Task.objects.bulk_create(
+            [
+                Task(
+                    run=0,
+                    depth=0,
+                    process=element_process,
+                    worker_run=worker_run,
+                    slug=f"unscheduled task {i}",
+                    state=State.Unscheduled,
+                ) for i in range(1, 4)
+            ]
+        )
+        task_1.parents.set([task_2])
+        task_3.parents.set([task_1, task_2])
         element.worker_run = worker_run
         element.worker_version = cls.worker_version
         element.save()

arkindex/sql_validation/corpus_delete.sql

@@ -192,6 +192,15 @@ DELETE
 FROM "training_dataset"
 WHERE "training_dataset"."corpus_id" = '{corpus_id}'::uuid;
 
+DELETE
+FROM "ponos_task_parents"
+WHERE "ponos_task_parents"."id" IN
+        (SELECT U0."id"
+         FROM "ponos_task_parents" U0
+         INNER JOIN "ponos_task" U1 ON (U0."from_task_id" = U1."id")
+         INNER JOIN "process_process" U2 ON (U1."process_id" = U2."id")
+         WHERE U2."corpus_id" = '{corpus_id}'::uuid);
+
 DELETE
 FROM "ponos_task"
 WHERE "ponos_task"."id" IN

arkindex/sql_validation/corpus_delete_top_level_type.sql

@@ -196,6 +196,15 @@ DELETE
 FROM "training_dataset"
 WHERE "training_dataset"."corpus_id" = '{corpus_id}'::uuid;
 
+DELETE
+FROM "ponos_task_parents"
+WHERE "ponos_task_parents"."id" IN
+        (SELECT U0."id"
+         FROM "ponos_task_parents" U0
+         INNER JOIN "ponos_task" U1 ON (U0."from_task_id" = U1."id")
+         INNER JOIN "process_process" U2 ON (U1."process_id" = U2."id")
+         WHERE U2."corpus_id" = '{corpus_id}'::uuid);
+
 DELETE
 FROM "ponos_task"
 WHERE "ponos_task"."id" IN