arkindex/documents/tasks.py

@@ -21,7 +21,7 @@ from arkindex.documents.models import (
     Transcription,
     TranscriptionEntity,
 )
-from arkindex.ponos.models import Task
+from arkindex.ponos.models import Artifact, Task
 from arkindex.process.models import Process, ProcessDatasetSet, ProcessElement, WorkerActivity, WorkerRun
 from arkindex.training.models import DatasetElement, DatasetSet
 from arkindex.users.models import User
@@ -41,6 +41,10 @@ def corpus_delete(corpus_id: str) -> None:
         corpus.top_level_type_id = None
         corpus.save(update_fields=["top_level_type_id"])
 
+    # Set process.element to NULL in all processes, to avoid IntegrityErrors when deleting elements, as
+    # elements are deleted before processes.
+    Process.objects.filter(corpus_id=corpus_id).exclude(element_id=None).update(element_id=None)
+
     # Delete all related objects, bypassing RESTRICTs deletion rules
     # and Django's way of loading everything into memory before deleting.
     querysets = [
@@ -67,6 +71,8 @@ def corpus_delete(corpus_id: str) -> None:
         Selection.objects.filter(element__corpus_id=corpus_id),
         corpus.memberships.all(),
         corpus.exports.all(),
+        # Delete task artifacts
+        Artifact.objects.filter(task__process__corpus_id=corpus_id),
         # ProcessDatasetSet M2M
         ProcessDatasetSet.objects.filter(set__dataset__corpus_id=corpus_id),
         ProcessDatasetSet.objects.filter(process__corpus_id=corpus_id),

arkindex/sql_validation/corpus_delete.sql

@@ -10,6 +10,11 @@ FROM "documents_corpus"
 WHERE "documents_corpus"."id" = '{corpus_id}'::uuid
 LIMIT 21;
 
+UPDATE "process_process"
+SET "element_id" = NULL
+WHERE ("process_process"."corpus_id" = '{corpus_id}'::uuid
+       AND NOT ("process_process"."element_id" IS NULL));
+
 SELECT "django_content_type"."id",
        "django_content_type"."app_label",
        "django_content_type"."model"
@@ -152,6 +157,15 @@ DELETE
 FROM "documents_corpusexport"
 WHERE "documents_corpusexport"."corpus_id" = '{corpus_id}'::uuid;
 
+DELETE
+FROM "ponos_artifact"
+WHERE "ponos_artifact"."id" IN
+        (SELECT U0."id"
+         FROM "ponos_artifact" U0
+         INNER JOIN "ponos_task" U1 ON (U0."task_id" = U1."id")
+         INNER JOIN "process_process" U2 ON (U1."process_id" = U2."id")
+         WHERE U2."corpus_id" = '{corpus_id}'::uuid);
+
 DELETE
 FROM "process_processdatasetset"
 WHERE "process_processdatasetset"."id" IN

arkindex/sql_validation/corpus_delete_top_level_type.sql

@@ -14,6 +14,11 @@ UPDATE "documents_corpus"
 SET "top_level_type_id" = NULL
 WHERE "documents_corpus"."id" = '{corpus_id}'::uuid;
 
+UPDATE "process_process"
+SET "element_id" = NULL
+WHERE ("process_process"."corpus_id" = '{corpus_id}'::uuid
+       AND NOT ("process_process"."element_id" IS NULL));
+
 SELECT "django_content_type"."id",
        "django_content_type"."app_label",
        "django_content_type"."model"
@@ -156,6 +161,15 @@ DELETE
 FROM "documents_corpusexport"
 WHERE "documents_corpusexport"."corpus_id" = '{corpus_id}'::uuid;
 
+DELETE
+FROM "ponos_artifact"
+WHERE "ponos_artifact"."id" IN
+        (SELECT U0."id"
+         FROM "ponos_artifact" U0
+         INNER JOIN "ponos_task" U1 ON (U0."task_id" = U1."id")
+         INNER JOIN "process_process" U2 ON (U1."process_id" = U2."id")
+         WHERE U2."corpus_id" = '{corpus_id}'::uuid);
+
 DELETE
 FROM "process_processdatasetset"
 WHERE "process_processdatasetset"."id" IN