Fix: finish manual and bulk zone OCR runs like other bulk edits

Runs reindexed each document separately, never sent document_updated
and did not bump modified for custom field writes. They now finish
through bulk_update_documents for the documents that got a value.
This commit is contained in:
stumpylog committed 2026-11-19 14:18:30 -08:00
1 parent 0bca50588d
commit 54be9dc9c5
9 files changed
+368 -45

No files matched your search

+74 -13
View File
@@ -14,18 +14,21 @@ from django.utils import timezone
from PIL import Image
from rest_framework import status
from documents import bulk_edit
from documents.models import CustomField
from documents.models import CustomFieldInstance
from documents.models import Document
from documents.models import MatchingModel
from documents.models import OcrTemplate
from documents.models import OcrTemplateZone
from documents.models import PaperlessTask
from documents.parsers import ParseError
from documents.search import get_backend
from documents.signals import document_consumption_finished
from documents.tasks import extract_zones
from documents.tasks import bulk_update_documents
from documents.zone_ocr import ZoneExtraction
from documents.zone_ocr import ZoneOcrReason
from documents.zone_ocr import ZoneOcrResult
from paperless_testing.factories import CustomFieldFactory
from paperless_testing.factories import DocumentFactory
from paperless_testing.factories import OcrTemplateFactory
@@ -148,6 +151,13 @@ def mock_extract(mocker: MockerFixture) -> MagicMock:
return mocker.patch("documents.zone_ocr.run_zone_extraction")
@pytest.fixture
def mock_finish(mocker: MockerFixture) -> MagicMock:
return mocker.patch(
"documents.views.bulk_edit.bulk_update_documents.apply_async",
)
@pytest.fixture
def mock_preview(mocker: MockerFixture) -> MagicMock:
return mocker.patch("documents.zone_ocr.extract_zone_preview")
@@ -232,6 +242,7 @@ class TestZoneOcrIndexing:
admin_client: APIClient,
django_capture_on_commit_callbacks: DjangoCaptureOnCommitCallbacks,
mock_zone_ocr_run: MagicMock,
mocker: MockerFixture,
) -> None:
"""
GIVEN:
@@ -305,6 +316,11 @@ class TestZoneOcrIndexing:
backend.add_or_update(document)
assert backend.search_ids("title:before", user=None) == [document.pk]
# The single-document finish is a queued task, run it inline here.
mocker.patch(
"documents.views.bulk_edit.bulk_update_documents.apply_async",
side_effect=lambda kwargs, **_: bulk_update_documents(**kwargs),
)
with django_capture_on_commit_callbacks(execute=True):
if entry_point == "ingestion":
document_consumption_finished.send(sender=None, document=file_document)
@@ -314,7 +330,7 @@ class TestZoneOcrIndexing:
)
assert response.status_code == status.HTTP_200_OK
else:
extract_zones(document.pk, template.pk)
bulk_edit.run_zone_ocr([document.pk], template=template.pk)
document.refresh_from_db()
assert document.title == "After OCR"
@@ -1801,7 +1817,7 @@ class TestOcrTemplateDelete:
@pytest.mark.django_db
@pytest.mark.usefixtures("paperless_dirs")
@pytest.mark.usefixtures("paperless_dirs", "mock_finish")
class TestRunZoneOcr:
"""Running zone OCR on documents through the run-zone-ocr endpoint."""
@@ -1873,11 +1889,60 @@ class TestRunZoneOcr:
assert response.status_code == status.HTTP_200_OK
assert response.json()["status"] == "completed"
mock_extract.assert_called_once_with(
document,
templates=[template],
reindex=True,
mock_extract.assert_called_once_with(document, templates=[template])
@pytest.mark.parametrize(
("reason", "finishes"),
[
pytest.param(None, True, id="written"),
pytest.param(ZoneOcrReason.NO_VALUE, False, id="nothing-written"),
],
)
def test_finishes_the_edit_only_when_a_value_was_written(
self,
admin_client: APIClient,
mock_extract: MagicMock,
mock_finish: MagicMock,
reason: ZoneOcrReason | None,
finishes: bool, # noqa: FBT001
) -> None:
"""
GIVEN:
- A document with a version and a template whose run does or does
not write a value
WHEN:
- Zone OCR is run on the version through the detail endpoint
THEN:
- The shared document update task is queued once for the root
document only when a value was written
- No per-document search reindex is requested directly
"""
document = DocumentFactory(content=CONTENT)
version = DocumentFactory(
root_document=document,
version_index=1,
content=CONTENT,
)
version.source_path.write_bytes(b"test PDF")
template = OcrTemplateFactory(name="Northwind", match="Northwind")
mock_extract.return_value = [
ZoneOcrResult(template="t", zone="z", custom_field="", reason=reason),
]
response = admin_client.post(
f"{RUN_ZONE_OCR_URL.format(pk=document.pk)}?version={version.pk}",
{"template": template.pk},
format="json",
)
assert response.status_code == status.HTTP_200_OK
if finishes:
mock_finish.assert_called_once_with(
kwargs={"document_ids": [document.pk]},
headers={"trigger_source": PaperlessTask.TriggerSource.SYSTEM},
)
else:
mock_finish.assert_not_called()
@pytest.mark.parametrize(
"version",
@@ -1923,11 +1988,7 @@ class TestRunZoneOcr:
assert response.status_code == status.HTTP_200_OK
assert response.json()["status"] == "completed"
mock_extract.assert_called_once_with(
file_document,
templates=[template],
reindex=True,
)
mock_extract.assert_called_once_with(file_document, templates=[template])
@pytest.mark.parametrize(
"version",
@@ -2190,7 +2251,7 @@ class TestRunZoneOcr:
@pytest.mark.django_db
@pytest.mark.usefixtures("paperless_dirs")
@pytest.mark.usefixtures("paperless_dirs", "mock_finish")
class TestRunZoneOcrPermissions:
"""Permission checks for the run-zone-ocr endpoint."""