Exclude manual results from deletion

arkindex/documents/consumers.py

@@ -114,8 +114,17 @@ class MLResultsConsumer(SyncConsumer):
 
         if element_id:
             logger.info('Deleting ML results on element {}'.format(element_id))
-            # This element and its direct children
-            elements = Element.objects.filter(Q(id=element_id) | Q(paths__path__last=element_id))
+            element = Element.objects.get(id=element_id)
+            if element.type.folder:
+                # The folder AND its children
+                elements = Element.objects.filter(id=element_id).values('id').union(
+                    # Disable ordering here because we do not need it and it adds an extra column,
+                    # causing the UNION to fail
+                    Element.objects.get_descending(element_id).order_by().values('id')
+                )
+            else:
+                elements = [element]
+
             if not corpus_id:
                 # The corpus ID is still used in some deletions; deduce it from the element.
                 corpus_id = Element.objects.get(id=element_id).corpus_id
@@ -124,8 +133,8 @@ class MLResultsConsumer(SyncConsumer):
             elements = Element.objects.filter(corpus_id=corpus_id)
 
         # Simple deletions for classifications and transcriptions.
-        _delete_queryset(Classification.objects.filter(element__in=elements), batch_size)
-        _delete_queryset(Transcription.objects.filter(element__in=elements), batch_size)
+        _delete_queryset(Classification.objects.filter(element__in=elements).exclude(source__slug='manual'), batch_size)
+        _delete_queryset(Transcription.objects.filter(element__in=elements).exclude(source__slug='manual'), batch_size)
 
         # Entity deletion is complex: they can be linked to different elements both on transcriptions and metadata.
         # Metadata are not considered ML results so we need to keep them: update them to unlink entities.

arkindex/documents/tests/consumers/test_ml_results_consumer.py

 from unittest.mock import patch
 from mock import AsyncMock
-from django.db.models import Q
 from django.test import override_settings
 from arkindex_common.enums import TranscriptionType, MetaType, EntityType
 from arkindex.project.tests import FixtureTestCase
@@ -130,13 +129,17 @@ class TestMLResultsConsumer(FixtureTestCase):
         self.assertFalse(get_layer_mock().send.called)
 
     @patch('arkindex.project.triggers.get_channel_layer')
-    def test_delete_element(self, get_layer_mock):
+    def test_delete_folder(self, get_layer_mock):
         get_layer_mock.return_value.send = AsyncMock()
         folder1_querysets = self._get_querysets(
-            Element.objects.filter(Q(id=self.folder1.id) | Q(paths__path__last=self.folder1.id))
+            Element.objects.filter(id=self.folder1.id).values('id').union(
+                Element.objects.get_descending(self.folder1.id).order_by().values('id')
+            )
         )
         folder2_querysets = self._get_querysets(
-            Element.objects.filter(Q(id=self.folder2.id) | Q(paths__path__last=self.folder2.id))
+            Element.objects.filter(id=self.folder2.id).values('id').union(
+                Element.objects.get_descending(self.folder2.id).order_by().values('id')
+            )
         )
 
         for queryset in folder1_querysets:
@@ -144,7 +147,7 @@ class TestMLResultsConsumer(FixtureTestCase):
         for queryset in folder2_querysets:
             self.assertTrue(queryset.exists())
 
-        with self.assertNumQueries(14):
+        with self.assertNumQueries(16):
             MLResultsConsumer({}).ml_results_delete({'element_id': str(self.folder1.id)})
 
         for queryset in folder1_querysets:
@@ -172,6 +175,51 @@ class TestMLResultsConsumer(FixtureTestCase):
             'drop': False,
         })
 
+    @patch('arkindex.project.triggers.get_channel_layer')
+    def test_delete_page(self, get_layer_mock):
+        get_layer_mock.return_value.send = AsyncMock()
+        page1_querysets = self._get_querysets(
+            Element.objects.filter(id=self.page1.id).values('id')
+        )
+        folder2_querysets = self._get_querysets(
+            Element.objects.filter(id=self.folder2.id).values('id').union(
+                Element.objects.get_descending(self.folder2.id).order_by().values('id')
+            )
+        )
+
+        for queryset in page1_querysets:
+            self.assertTrue(queryset.exists())
+        for queryset in folder2_querysets:
+            self.assertTrue(queryset.exists())
+
+        with self.assertNumQueries(16):
+            MLResultsConsumer({}).ml_results_delete({'element_id': str(self.page1.id)})
+
+        for queryset in page1_querysets:
+            self.assertFalse(queryset.exists())
+        for queryset in folder2_querysets:  # This folder should not change
+            self.assertTrue(queryset.exists())
+
+        self.assertEqual(self.page1.metadatas.count(), 2)
+        self.assertEqual(self.page2.metadatas.count(), 2)
+        metadata = self.page1.metadatas.get(value='Some entity 1')
+        self.assertEqual(metadata.type, MetaType.Text)
+        self.assertIsNone(metadata.entity)
+        metadata = self.page2.metadatas.get(value='Some entity 2')
+        self.assertEqual(metadata.type, MetaType.Text)
+        self.assertIsNotNone(metadata.entity)
+
+        get_layer_mock().send.assert_called_once_with('reindex', {
+            'type': 'reindex.start',
+            'element': str(self.page1.id),
+            'corpus': str(self.corpus.id),
+            'entity': None,
+            'transcriptions': True,
+            'elements': True,
+            'entities': True,
+            'drop': False,
+        })
+
     @override_settings(ARKINDEX_FEATURES={'search': False})
     @patch('arkindex.project.triggers.get_channel_layer')
     def test_delete_element_no_search(self, get_layer_mock):