diff --git a/arkindex/documents/tasks.py b/arkindex/documents/tasks.py
index ce4c680ae9d483e6b94da0595336902701ed66c0..80ece7ecc1462d39da756ee716480aa7df48747a 100644
--- a/arkindex/documents/tasks.py
+++ b/arkindex/documents/tasks.py
@@ -16,6 +16,7 @@ from arkindex.documents.models import (
Corpus,
Element,
ElementPath,
+ Entity,
MetaData,
Selection,
Transcription,
@@ -176,6 +177,7 @@ def worker_results_delete(
Q(transcription__element__corpus_id=corpus_id)
| Q(entity__corpus_id=corpus_id)
)
+ entities = Entity.objects.filter(corpus_id=corpus_id)
metadata = MetaData.objects.filter(element__corpus_id=corpus_id)
worker_activities = WorkerActivity.objects.filter(element__corpus_id=corpus_id)
@@ -186,6 +188,7 @@ def worker_results_delete(
transcriptions = transcriptions.filter(worker_run_id=worker_run_id)
transcription_entities = transcription_entities.filter(transcription__worker_run_id=worker_run_id)
worker_transcription_entities = worker_transcription_entities.filter(worker_run_id=worker_run_id)
+ entities = entities.filter(worker_run_id=worker_run_id)
metadata = metadata.filter(worker_run_id=worker_run_id)
# There is no worker_run_id on Worker Activities so the best thing we can do is delete the worker activities
# attached to the elements produced with that worker run, and they are already being deleted by elements.trash()
@@ -197,6 +200,7 @@ def worker_results_delete(
transcriptions = transcriptions.filter(worker_version_id=version_id)
transcription_entities = transcription_entities.filter(transcription__worker_version_id=version_id)
worker_transcription_entities = worker_transcription_entities.filter(worker_version_id=version_id)
+ entities = entities.filter(worker_version_id=version_id)
metadata = metadata.filter(worker_version_id=version_id)
worker_activities = worker_activities.filter(worker_version_id=version_id)
# Otherwise, select everything that has any worker version ID. (When something has been created
@@ -211,9 +215,14 @@ def worker_results_delete(
transcriptions = transcriptions.exclude(worker_version_id=None)
transcription_entities = transcription_entities.exclude(transcription__worker_version_id=None)
worker_transcription_entities = worker_transcription_entities.exclude(worker_version_id=None)
+ entities = entities.exclude(worker_version_id=None)
metadata = metadata.exclude(worker_version_id=None)
if element_id:
+ # When running with a parent element, we do not delete any entities,
+ # as checking whether they are used in other elements is too costly.
+ entities = entities.none()
+
# include_children causes a deletion *only* on the element's descendants.
# To also delete on the element itself, we will call this task synchronously with the same arguments and include_children=False.
# This is used to avoid filtering by Q(id=element_id) | Q(paths__path__overlap=[element_id]) all at once,
@@ -245,6 +254,7 @@ def worker_results_delete(
transcriptions = transcriptions.filter(worker_run__model_version_id=model_version_id)
transcription_entities = transcription_entities.filter(transcription__worker_run__model_version_id=model_version_id)
worker_transcription_entities = worker_transcription_entities.filter(worker_run__model_version_id=model_version_id)
+ entities = entities.filter(worker_run__model_version_id=model_version_id)
metadata = metadata.filter(worker_run__model_version_id=model_version_id)
worker_activities = worker_activities.filter(model_version_id=model_version_id)
@@ -256,6 +266,7 @@ def worker_results_delete(
transcriptions = transcriptions.filter(worker_run__configuration_id=None)
transcription_entities = transcription_entities.filter(transcription__worker_run__configuration_id=None)
worker_transcription_entities = worker_transcription_entities.filter(worker_run__configuration_id=None)
+ entities = entities.filter(worker_run__configuration_id=None)
metadata = metadata.filter(worker_run__configuration_id=None)
worker_activities = worker_activities.filter(configuration_id=None)
else:
@@ -264,16 +275,26 @@ def worker_results_delete(
transcriptions = transcriptions.filter(worker_run__configuration_id=configuration_id)
transcription_entities = transcription_entities.filter(transcription__worker_run__configuration_id=configuration_id)
worker_transcription_entities = worker_transcription_entities.filter(worker_run__configuration_id=configuration_id)
+ entities = entities.filter(worker_run__configuration_id=configuration_id)
metadata = metadata.filter(worker_run__configuration_id=configuration_id)
worker_activities = worker_activities.filter(configuration_id=configuration_id)
elements.trash()
classifications.delete()
+
# Delete TranscriptionEntities before transcriptions so that we can delete transcriptions using a single DELETE query
transcription_entities.delete()
worker_transcription_entities.delete()
transcriptions._raw_delete(using="default")
+
metadata.delete()
+
+ # Remove entities from metadata and remove their associated transcription entities first
+ # so that entities can be deleted in one query.
+ MetaData.objects.filter(entity__in=entities).update(entity_id=None)
+ TranscriptionEntity.objects.filter(entity__in=entities).delete()
+ entities._raw_delete(using="default")
+
worker_activities.delete()
if element_id and include_children:
diff --git a/arkindex/documents/tests/tasks/test_worker_results_delete.py b/arkindex/documents/tests/tasks/test_worker_results_delete.py
index 52546e1bac401f5cd53028fb17a460e07c755b31..2906bc25b49fd5fdeb1c5ff5ff85b51ad33f4bfa 100644
--- a/arkindex/documents/tests/tasks/test_worker_results_delete.py
+++ b/arkindex/documents/tests/tasks/test_worker_results_delete.py
@@ -211,6 +211,7 @@ class TestDeleteWorkerResults(FixtureTestCase):
self.transcription2,
self.transcription_entity1,
self.transcription_entity2,
+ self.entity,
)
def test_run_no_configuration_filter(self):
@@ -261,6 +262,7 @@ class TestDeleteWorkerResults(FixtureTestCase):
self.transcription_entity1,
self.transcription_entity2,
self.page2,
+ self.entity,
)
def test_run_dataset_failure(self):
diff --git a/arkindex/sql_validation/worker_results_delete_all_versions.sql b/arkindex/sql_validation/worker_results_delete_all_versions.sql
index afbb1ee13b353ad86d4a0a1746799073eaf36f97..dc5367c8ac66ad481b25b1a74e09136a81229312 100644
--- a/arkindex/sql_validation/worker_results_delete_all_versions.sql
+++ b/arkindex/sql_validation/worker_results_delete_all_versions.sql
@@ -268,9 +268,31 @@ WHERE "documents_metadata"."id" IN
WHERE (U1."corpus_id" = '{corpus_id}'::uuid
AND NOT (U0."worker_version_id" IS NULL)));
+UPDATE "documents_metadata"
+SET "entity_id" = NULL
+WHERE "documents_metadata"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND NOT (U0."worker_version_id" IS NULL)));
+
+DELETE
+FROM "documents_transcriptionentity"
+WHERE "documents_transcriptionentity"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND NOT (U0."worker_version_id" IS NULL)));
+
+DELETE
+FROM "documents_entity"
+WHERE ("documents_entity"."corpus_id" = '{corpus_id}'::uuid
+ AND NOT ("documents_entity"."worker_version_id" IS NULL));
+
DELETE
FROM "process_workeractivity"
-WHERE "process_workeractivity"."id" IN (SELECT U0."id"
+WHERE "process_workeractivity"."id" IN
+ (SELECT U0."id"
FROM "process_workeractivity" U0
INNER JOIN "documents_element" U1 ON (U0."element_id" = U1."id")
WHERE U1."corpus_id" = '{corpus_id}'::uuid)
diff --git a/arkindex/sql_validation/worker_results_delete_configuration_filter.sql b/arkindex/sql_validation/worker_results_delete_configuration_filter.sql
index f1e2ed0627340a7c672b6d184bfb4071c1defcab..9ee04a8506fbc707182614c5ef055eeb275b63ff 100644
--- a/arkindex/sql_validation/worker_results_delete_configuration_filter.sql
+++ b/arkindex/sql_validation/worker_results_delete_configuration_filter.sql
@@ -175,6 +175,36 @@ WHERE "documents_metadata"."id" IN
AND NOT (U0."worker_version_id" IS NULL)
AND U4."configuration_id" = '{configuration_id}'::uuid));
+UPDATE "documents_metadata"
+SET "entity_id" = NULL
+WHERE "documents_metadata"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ INNER JOIN "process_workerrun" U3 ON (U0."worker_run_id" = U3."id")
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND NOT (U0."worker_version_id" IS NULL)
+ AND U3."configuration_id" = '{configuration_id}'::uuid));
+
+DELETE
+FROM "documents_transcriptionentity"
+WHERE "documents_transcriptionentity"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ INNER JOIN "process_workerrun" U3 ON (U0."worker_run_id" = U3."id")
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND NOT (U0."worker_version_id" IS NULL)
+ AND U3."configuration_id" = '{configuration_id}'::uuid));
+
+DELETE
+FROM "documents_entity"
+WHERE "documents_entity"."id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ INNER JOIN "process_workerrun" U3 ON (U0."worker_run_id" = U3."id")
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND NOT (U0."worker_version_id" IS NULL)
+ AND U3."configuration_id" = '{configuration_id}'::uuid));
+
DELETE
FROM "process_workeractivity"
WHERE "process_workeractivity"."id" IN
diff --git a/arkindex/sql_validation/worker_results_delete_in_corpus.sql b/arkindex/sql_validation/worker_results_delete_in_corpus.sql
index 3f7e78812071e184a6aa9a62d332b70dd4751789..e9f17f739dcf1e1ca20bdef82b5fceb45b731ee8 100644
--- a/arkindex/sql_validation/worker_results_delete_in_corpus.sql
+++ b/arkindex/sql_validation/worker_results_delete_in_corpus.sql
@@ -268,6 +268,27 @@ WHERE "documents_metadata"."id" IN
WHERE (U1."corpus_id" = '{corpus_id}'::uuid
AND U0."worker_version_id" = '{version_id}'::uuid));
+UPDATE "documents_metadata"
+SET "entity_id" = NULL
+WHERE "documents_metadata"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND U0."worker_version_id" = '{version_id}'::uuid));
+
+DELETE
+FROM "documents_transcriptionentity"
+WHERE "documents_transcriptionentity"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND U0."worker_version_id" = '{version_id}'::uuid));
+
+DELETE
+FROM "documents_entity"
+WHERE ("documents_entity"."corpus_id" = '{corpus_id}'::uuid
+ AND "documents_entity"."worker_version_id" = '{version_id}'::uuid);
+
DELETE
FROM "process_workeractivity"
WHERE "process_workeractivity"."id" IN
diff --git a/arkindex/sql_validation/worker_results_delete_in_corpus_worker_run.sql b/arkindex/sql_validation/worker_results_delete_in_corpus_worker_run.sql
index e01a89ead286819eb18017f01af71bf8e4960177..dcb016608658431feb76d7de79fc2b3918b6a8cb 100644
--- a/arkindex/sql_validation/worker_results_delete_in_corpus_worker_run.sql
+++ b/arkindex/sql_validation/worker_results_delete_in_corpus_worker_run.sql
@@ -141,4 +141,25 @@ WHERE "documents_metadata"."id" IN
FROM "documents_metadata" U0
INNER JOIN "documents_element" U1 ON (U0."element_id" = U1."id")
WHERE (U1."corpus_id" = '{corpus_id}'::uuid
- AND U0."worker_run_id" = '{worker_run_id}'::uuid))
+ AND U0."worker_run_id" = '{worker_run_id}'::uuid));
+
+UPDATE "documents_metadata"
+SET "entity_id" = NULL
+WHERE "documents_metadata"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND U0."worker_run_id" = '{worker_run_id}'::uuid));
+
+DELETE
+FROM "documents_transcriptionentity"
+WHERE "documents_transcriptionentity"."entity_id" IN
+ (SELECT U0."id"
+ FROM "documents_entity" U0
+ WHERE (U0."corpus_id" = '{corpus_id}'::uuid
+ AND U0."worker_run_id" = '{worker_run_id}'::uuid));
+
+DELETE
+FROM "documents_entity"
+WHERE ("documents_entity"."corpus_id" = '{corpus_id}'::uuid
+ AND "documents_entity"."worker_run_id" = '{worker_run_id}'::uuid)