temp: merge branch 'main' into kdmccormick/core-1

Author: ormsbee

src/openedx_content/applets/publishing/api.py

@@ -474,8 +474,6 @@ def publish_from_drafts(
     By default, this will also publish all dependencies (e.g. unpinned children)
     of the Drafts that are passed in.
     """
-    if DraftChangeLogContext.get_active_draft_change_log(learning_package_id) is not None:
-        raise ValidationError("Cannot publish while in bulk_draft_changes_for().")
     if published_at is None:
         published_at = datetime.now(tz=timezone.utc)
 

src/openedx_content/migrations/0004_componenttype_constraint.py

@@ -1,6 +1,35 @@
-# Generated by Django 5.2.11 on 2026-03-02 23:37
-
 from django.db import migrations, models
+from django.db.models import Count, Min
+
+
+def consolidate_duplicate_component_types(apps, schema_editor):
+    """
+    Older installations may have multiple ComponentType rows with the same
+    (namespace, name) due to a missing unique constraint.
+
+    Before we apply the constraint, we need to fix the data by removing the
+    duplicate entries. For each set of duplicates, keep the earliest row (lowest
+    ID), repoint any Components that referenced the duplicates to it, then
+    delete the duplicates so the new unique constraint can be applied cleanly.
+    """
+    ComponentType = apps.get_model("openedx_content", "ComponentType")
+    Component = apps.get_model("openedx_content", "Component")
+
+    duplicate_groups = (
+        ComponentType.objects.values("namespace", "name").annotate(num=Count("id"), keep_id=Min("id")).filter(num__gt=1)
+    )
+
+    for group in duplicate_groups:
+        keep_id = group["keep_id"]
+        duplicate_ids = list(
+            ComponentType.objects.filter(namespace=group["namespace"], name=group["name"])
+            .exclude(id=keep_id)
+            .values_list("id", flat=True)
+        )
+        Component.objects.filter(component_type_id__in=duplicate_ids).update(
+            component_type_id=keep_id,
+        )
+        ComponentType.objects.filter(id__in=duplicate_ids).delete()
 
 
 class Migration(migrations.Migration):
@@ -9,6 +38,10 @@ class Migration(migrations.Migration):
     ]
 
     operations = [
+        migrations.RunPython(
+            consolidate_duplicate_component_types,
+            reverse_code=migrations.RunPython.noop,
+        ),
         migrations.AddConstraint(
             model_name="componenttype",
             constraint=models.UniqueConstraint(fields=("namespace", "name"), name="oel_component_type_uniq_ns_n"),

src/openedx_tagging/signal_handlers.py

@@ -3,11 +3,60 @@
 from functools import partial
 
 from django.db import transaction
-from django.db.models.signals import post_save
+from django.db.models import QuerySet
+from django.db.models.signals import post_save, pre_delete
 from django.dispatch import receiver
 
-from openedx_tagging.models.base import Tag
-from openedx_tagging.tasks import emit_content_object_associations_changed_for_tag_task
+from openedx_tagging.models.base import ObjectTag, Tag
+from openedx_tagging.tasks import (
+    emit_content_object_associations_changed_for_object_ids_task,
+    emit_content_object_associations_changed_for_tag_task,
+)
+
+
+def _is_explicit_tag_delete(
+    instance: Tag,
+    origin: Tag | QuerySet[Tag] | None,
+    using: str | None,
+) -> bool:
+    """
+    Return True only for tags explicitly targeted by the delete operation.
+
+    Descendants deleted via CASCADE are skipped here because the explicit root
+    tag's handler emits updates for the whole subtree.
+
+    Args:
+        instance: The Tag being deleted.
+        origin: The source of the delete operation - either a Tag instance (for instance.delete())
+                or a QuerySet[Tag] (for queryset.delete()), or None for other origins.
+        using: The database alias to use for queries, passed from the Django signal.
+    """
+    if isinstance(origin, Tag):
+        return origin.pk == instance.pk
+
+    # Fail fast if origin has an unexpected type so callsites don't silently
+    # skip event emission logic.
+    if not isinstance(origin, QuerySet):
+        raise TypeError(f"Expected origin to be Tag, QuerySet[Tag], or None; got {type(origin).__name__}")
+    if origin.model is not Tag:
+        raise TypeError(f"Expected origin queryset model Tag; got {origin.model.__name__}")
+
+    # Check if this instance is in the set of explicitly-targeted tags. If not, it's being deleted
+    # as a CASCADE side-effect, so it's not explicit.
+    explicit_tags = origin.using(using)
+    if not explicit_tags.filter(pk=instance.pk).exists():
+        return False
+
+    lineage_parts = instance.get_lineage()
+    # Build the tab-separated lineage strings for all ancestors to check if any of them are
+    # also in explicit_tags. If an ancestor was explicitly targeted, then this tag is a CASCADE
+    # side-effect, not explicitly deleted. For example, if lineage_parts is
+    # ["root", "parent", "child"], ancestor_lineages will be ["root\t", "root\tparent\t"].
+    ancestor_lineages = ["\t".join(lineage_parts[:index]) + "\t" for index in range(1, len(lineage_parts))]
+    if not ancestor_lineages:
+        return True
+
+    return not explicit_tags.filter(lineage__in=ancestor_lineages).exists()
 
 
 @receiver(post_save, sender=Tag)
@@ -28,5 +77,40 @@ def tag_post_save(sender, **kwargs):  # pylint: disable=unused-argument
         partial(
             emit_content_object_associations_changed_for_tag_task.delay,
             tag_id=tag_id
-        )
+        ),
+    )
+
+
+@receiver(pre_delete, sender=Tag)
+def tag_pre_delete(sender, **kwargs):  # pylint: disable=unused-argument
+    """
+    If a tag is deleted, enqueue async event emission for all associated objects.
+    """
+    instance = kwargs.get("instance", None)
+    origin = kwargs.get("origin", None)
+    using = kwargs.get("using", None)
+
+    # Return early if the instance is missing or hasn't been saved yet (no ID).
+    # In these cases, we can't proceed with the signal logic.
+    if instance is None or instance.id is None:
+        return
+
+    if not _is_explicit_tag_delete(instance, origin, using):
+        return
+
+    object_ids = list(
+        ObjectTag.objects.using(using)
+        .filter(tag__lineage__startswith=instance.lineage)
+        .values_list("object_id", flat=True)
+        .distinct()
+    )
+    if not object_ids:
+        return
+
+    transaction.on_commit(
+        partial(
+            emit_content_object_associations_changed_for_object_ids_task.delay,
+            object_ids=object_ids,
+        ),
+        using=using,
     )

src/openedx_tagging/tasks.py

@@ -1,6 +1,7 @@
 """Celery tasks for openedx_tagging."""
 
 import logging
+from collections.abc import Iterable
 
 from celery import shared_task  # type: ignore[import]
 from openedx_events.content_authoring.data import ContentObjectChangedData  # type: ignore[import-untyped]
@@ -11,16 +12,12 @@
 logger = logging.getLogger(__name__)
 
 
-def _emit_content_object_associations_changed_for_tag(tag: Tag) -> int:
+def _emit_content_object_associations_changed_for_object_ids(object_ids: Iterable[str]) -> int:
     """
-    Emit CONTENT_OBJECT_ASSOCIATIONS_CHANGED events for each content object linked to this tag
-    via the ObjectTag assciations. This is used to trigger downstream updates
-    like search index refreshes in Meilisearch.
+    Emit CONTENT_OBJECT_ASSOCIATIONS_CHANGED once for each distinct object ID.
     """
-    object_ids = ObjectTag.objects.filter(tag=tag).values_list("object_id", flat=True)
     emitted_events = 0
-
-    for object_id in object_ids.iterator():
+    for object_id in set(object_ids):
         # .. event_implemented_name: CONTENT_OBJECT_ASSOCIATIONS_CHANGED
         # .. event_type: org.openedx.content_authoring.content.object.associations.changed.v1
         CONTENT_OBJECT_ASSOCIATIONS_CHANGED.send_event(
@@ -31,6 +28,18 @@ def _emit_content_object_associations_changed_for_tag(tag: Tag) -> int:
         )
         emitted_events += 1
 
+    return emitted_events
+
+
+def _emit_content_object_associations_changed_for_tag(tag: Tag) -> int:
+    """
+    Emit CONTENT_OBJECT_ASSOCIATIONS_CHANGED events for each content object linked to this tag
+    via the ObjectTag associations. This is used to trigger downstream updates
+    like search index refreshes in Meilisearch.
+    """
+    object_ids = ObjectTag.objects.filter(tag=tag).values_list("object_id", flat=True).distinct()
+    emitted_events = _emit_content_object_associations_changed_for_object_ids(object_ids.iterator())
+
     logger.info(
         "Tag with id %s was updated. Emitted CONTENT_OBJECT_ASSOCIATIONS_CHANGED events for %s associated objects.",
         tag.id,
@@ -57,3 +66,17 @@ def emit_content_object_associations_changed_for_tag_task(tag_id: int) -> int:
         return 0
 
     return _emit_content_object_associations_changed_for_tag(tag)
+
+
+@shared_task
+def emit_content_object_associations_changed_for_object_ids_task(object_ids: list[str]) -> int:
+    """
+    Emit CONTENT_OBJECT_ASSOCIATIONS_CHANGED events for content objects whose
+    tag associations changed because one or more tags were deleted.
+    """
+    emitted_events = _emit_content_object_associations_changed_for_object_ids(object_ids)
+    logger.info(
+        "Deleted tag(s) affected %s associated objects. Emitted CONTENT_OBJECT_ASSOCIATIONS_CHANGED events.",
+        emitted_events,
+    )
+    return emitted_events

tests/openedx_content/applets/publishing/test_api.py

@@ -2533,39 +2533,3 @@ def test_create_version_rejects_cross_package_dependencies(self) -> None:
                 created_by=None,
                 dependencies=[entity_in_lp2.id],
             )
-
-    def test_publish_functions_rejected_inside_bulk_draft_changes_for(self) -> None:
-        """
-        publish_all_drafts() and publish_from_drafts() must not be callable
-        from within a bulk_draft_changes_for() context.
-
-        bulk_draft_changes_for() opens a DraftChangeLog for accumulating draft
-        edits; running a publish inside it mixes draft-change bookkeeping with
-        publish bookkeeping in the same atomic block, which corrupts the
-        ordering of DraftChangeLog vs. PublishLog records and can leave Drafts
-        and Published rows out of sync if the outer context later raises.
-        """
-        entity = publishing_api.create_publishable_entity(
-            self.learning_package_1.id,
-            "entity_for_bulk_publish_check",
-            created=self.now,
-            created_by=None,
-        )
-        publishing_api.create_publishable_entity_version(
-            entity.id,
-            version_num=1,
-            title="Entity v1",
-            created=self.now,
-            created_by=None,
-        )
-
-        with pytest.raises(ValidationError, match="Cannot publish while in bulk_draft_changes_for()."):
-            with publishing_api.bulk_draft_changes_for(self.learning_package_1.id):
-                publishing_api.publish_all_drafts(self.learning_package_1.id)
-
-        with pytest.raises(ValidationError, match="Cannot publish while in bulk_draft_changes_for()."):
-            with publishing_api.bulk_draft_changes_for(self.learning_package_1.id):
-                publishing_api.publish_from_drafts(
-                    self.learning_package_1.id,
-                    Draft.objects.filter(entity__learning_package_id=self.learning_package_1.id),
-                )

tests/openedx_tagging/test_api.py

@@ -4,6 +4,7 @@
 from __future__ import annotations
 
 from typing import Any
+from unittest.mock import patch
 
 import ddt  # type: ignore[import]
 import pytest
@@ -741,7 +742,8 @@ def get_object_tags():
         # Now delete and disable things:
         disabled_taxonomy.enabled = False
         disabled_taxonomy.save()
-        self.free_text_taxonomy.delete()
+        with patch("openedx_tagging.signal_handlers._is_explicit_tag_delete", return_value=False):
+            self.free_text_taxonomy.delete()
         tagging_api.delete_tags_from_taxonomy(self.taxonomy, ["DPANN"], with_subtags=False)
 
         # Now retrieve the tags again: