feat: add language auto-tagging with feature flag (openedx#32907)

openedx/features/content_tagging/api.py

@@ -1,12 +1,13 @@
 """
 Content Tagging APIs
 """
-from typing import Iterator, List, Type, Union
+from __future__ import annotations
+
+from typing import Iterator, List, Type
 
 import openedx_tagging.core.tagging.api as oel_tagging
 from django.db.models import QuerySet
-from opaque_keys.edx.keys import LearningContextKey
-from opaque_keys.edx.locator import BlockUsageLocator
+from opaque_keys.edx.keys import CourseKey, UsageKey
 from openedx_tagging.core.tagging.models import Taxonomy
 from organizations.models import Organization
 
@@ -117,9 +118,9 @@ def get_content_tags(
 
 def tag_content_object(
     taxonomy: Taxonomy,
-    tags: List,
-    object_id: Union[BlockUsageLocator, LearningContextKey],
-) -> List[ContentObjectTag]:
+    tags: list,
+    object_id: CourseKey | UsageKey,
+) -> list[ContentObjectTag]:
     """
     This is the main API to use when you want to add/update/delete tags from a content object (e.g. an XBlock or
     course).
@@ -150,4 +151,5 @@ def tag_content_object(
 get_taxonomy = oel_tagging.get_taxonomy
 get_taxonomies = oel_tagging.get_taxonomies
 get_tags = oel_tagging.get_tags
+delete_object_tags = oel_tagging.delete_object_tags
 resync_object_tags = oel_tagging.resync_object_tags

openedx/features/content_tagging/apps.py

@@ -10,3 +10,7 @@ class ContentTaggingConfig(AppConfig):
 
     default_auto_field = "django.db.models.BigAutoField"
     name = "openedx.features.content_tagging"
+
+    def ready(self):
+        # Connect signal handlers
+        from . import handlers  # pylint: disable=unused-import

openedx/features/content_tagging/handlers.py

@@ -0,0 +1,76 @@
+"""
+Automatic tagging of content
+"""
+
+import logging
+
+from django.dispatch import receiver
+from openedx_events.content_authoring.data import CourseData, XBlockData
+from openedx_events.content_authoring.signals import COURSE_CREATED, XBLOCK_CREATED, XBLOCK_DELETED, XBLOCK_UPDATED
+
+from .tasks import delete_course_tags
+from .tasks import (
+    delete_xblock_tags,
+    update_course_tags,
+    update_xblock_tags
+)
+from .toggles import CONTENT_TAGGING_AUTO
+
+log = logging.getLogger(__name__)
+
+
+@receiver(COURSE_CREATED)
+def auto_tag_course(**kwargs):
+    """
+    Automatically tag course based on their metadata
+    """
+    course_data = kwargs.get("course", None)
+    if not course_data or not isinstance(course_data, CourseData):
+        log.error("Received null or incorrect data for event")
+        return
+
+    if not CONTENT_TAGGING_AUTO.is_enabled(course_data.course_key):
+        return
+
+    update_course_tags.delay(str(course_data.course_key))
+
+
+@receiver(XBLOCK_CREATED)
+@receiver(XBLOCK_UPDATED)
+def auto_tag_xblock(**kwargs):
+    """
+    Automatically tag XBlock based on their metadata
+    """
+    xblock_info = kwargs.get("xblock_info", None)
+    if not xblock_info or not isinstance(xblock_info, XBlockData):
+        log.error("Received null or incorrect data for event")
+        return
+
+    if not CONTENT_TAGGING_AUTO.is_enabled(xblock_info.usage_key.course_key):
+        return
+
+    if xblock_info.block_type == "course":
+        # Course update is handled by XBlock of course type
+        update_course_tags.delay(str(xblock_info.usage_key.course_key))
+
+    update_xblock_tags.delay(str(xblock_info.usage_key))
+
+
+@receiver(XBLOCK_DELETED)
+def delete_tag_xblock(**kwargs):
+    """
+    Automatically delete XBlock auto tags.
+    """
+    xblock_info = kwargs.get("xblock_info", None)
+    if not xblock_info or not isinstance(xblock_info, XBlockData):
+        log.error("Received null or incorrect data for event")
+        return
+
+    if not CONTENT_TAGGING_AUTO.is_enabled(xblock_info.usage_key.course_key):
+        return
+
+    if xblock_info.block_type == "course":
+        # Course deletion is handled by XBlock of course type
+        delete_course_tags.delay(str(xblock_info.usage_key.course_key))
+
+    delete_xblock_tags.delay(str(xblock_info.usage_key))

openedx/features/content_tagging/migrations/0003_system_defined_fixture.py

@@ -1,37 +1,62 @@
 # Generated by Django 3.2.20 on 2023-07-11 22:57
 
 from django.db import migrations
-from django.core.management import call_command
-from openedx.features.content_tagging.models import ContentLanguageTaxonomy
 
 
 def load_system_defined_taxonomies(apps, schema_editor):
     """
     Creates system defined taxonomies
-    """ 
+    """
 
     # Create system defined taxonomy instances
-    call_command('loaddata', '--app=content_tagging', 'system_defined.yaml')
+    Taxonomy = apps.get_model("oel_tagging", "Taxonomy")
+    author_taxonomy = Taxonomy(
+        pk=-2,
+        name="Content Authors",
+        description="Allows tags for any user ID created on the instance.",
+        enabled=True,
+        required=True,
+        allow_multiple=False,
+        allow_free_text=False,
+        visible_to_authors=False,
+    )
+    ContentAuthorTaxonomy = apps.get_model("content_tagging", "ContentAuthorTaxonomy")
+    author_taxonomy.taxonomy_class = ContentAuthorTaxonomy
+    author_taxonomy.save()
+
+    org_taxonomy = Taxonomy(
+        pk=-3,
+        name="Organizations",
+        description="Allows tags for any organization ID created on the instance.",
+        enabled=True,
+        required=True,
+        allow_multiple=False,
+        allow_free_text=False,
+        visible_to_authors=False,
+    )
+    ContentOrganizationTaxonomy = apps.get_model("content_tagging", "ContentOrganizationTaxonomy")
+    org_taxonomy.taxonomy_class = ContentOrganizationTaxonomy
+    org_taxonomy.save()
 
     # Adding taxonomy class to the language taxonomy
-    Taxonomy = apps.get_model('oel_tagging', 'Taxonomy')
     language_taxonomy = Taxonomy.objects.get(id=-1)
+    ContentLanguageTaxonomy = apps.get_model("content_tagging", "ContentLanguageTaxonomy")
     language_taxonomy.taxonomy_class = ContentLanguageTaxonomy
+    language_taxonomy.save()
 
 
 def revert_system_defined_taxonomies(apps, schema_editor):
     """
     Deletes all system defined taxonomies
     """
-    Taxonomy = apps.get_model('oel_tagging', 'Taxonomy')
+    Taxonomy = apps.get_model("oel_tagging", "Taxonomy")
     Taxonomy.objects.get(id=-2).delete()
     Taxonomy.objects.get(id=-3).delete()
 
 
 class Migration(migrations.Migration):
-
     dependencies = [
-        ('content_tagging', '0002_system_defined_taxonomies'),
+        ("content_tagging", "0002_system_defined_taxonomies"),
     ]
 
     operations = [

openedx/features/content_tagging/migrations/0004_system_defined_org.py

@@ -0,0 +1,49 @@
+from django.db import migrations
+
+
+def load_system_defined_org_taxonomies(apps, _schema_editor):
+    """
+    Associates the system defined taxonomy Language (id=-1) to all orgs and
+    removes the ContentOrganizationTaxonomy (id=-3) from the database
+    """
+    TaxonomyOrg = apps.get_model("content_tagging", "TaxonomyOrg")
+    TaxonomyOrg.objects.create(id=-1, taxonomy_id=-1, org=None)
+
+    Taxonomy = apps.get_model("oel_tagging", "Taxonomy")
+    Taxonomy.objects.get(id=-3).delete()
+
+
+
+
+def revert_system_defined_org_taxonomies(apps, _schema_editor):
+    """
+    Deletes association of system defined taxonomy Language (id=-1) to all orgs and
+    creates the ContentOrganizationTaxonomy (id=-3) in the database
+    """
+    TaxonomyOrg = apps.get_model("content_tagging", "TaxonomyOrg")
+    TaxonomyOrg.objects.get(id=-1).delete()
+
+    Taxonomy = apps.get_model("oel_tagging", "Taxonomy")
+    org_taxonomy = Taxonomy(
+        pk=-3,
+        name="Organizations",
+        description="Allows tags for any organization ID created on the instance.",
+        enabled=True,
+        required=True,
+        allow_multiple=False,
+        allow_free_text=False,
+        visible_to_authors=False,
+    )
+    ContentOrganizationTaxonomy = apps.get_model("content_tagging", "ContentOrganizationTaxonomy")
+    org_taxonomy.taxonomy_class = ContentOrganizationTaxonomy
+    org_taxonomy.save()
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("content_tagging", "0003_system_defined_fixture"),
+    ]
+
+    operations = [
+        migrations.RunPython(load_system_defined_org_taxonomies, revert_system_defined_org_taxonomies),
+    ]

openedx/features/content_tagging/migrations/0005_auto_20230830_1517.py

@@ -0,0 +1,19 @@
+# Generated by Django 3.2.20 on 2023-08-30 15:17
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('content_tagging', '0004_system_defined_org'),
+    ]
+
+    operations = [
+        migrations.DeleteModel(
+            name='ContentOrganizationTaxonomy',
+        ),
+        migrations.DeleteModel(
+            name='OrganizationModelObjectTag',
+        ),
+    ]

openedx/features/content_tagging/models/__init__.py

@@ -9,5 +9,4 @@
 from .system_defined import (
     ContentLanguageTaxonomy,
     ContentAuthorTaxonomy,
-    ContentOrganizationTaxonomy,
 )

openedx/features/content_tagging/models/base.py

@@ -1,7 +1,7 @@
 """
 Content Tagging models
 """
-from typing import List, Union
+from __future__ import annotations
 
 from django.db import models
 from django.db.models import Exists, OuterRef, Q, QuerySet
@@ -49,7 +49,7 @@ class Meta:
 
     @classmethod
     def get_relationships(
-        cls, taxonomy: Taxonomy, rel_type: RelType, org_short_name: Union[str, None] = None
+        cls, taxonomy: Taxonomy, rel_type: RelType, org_short_name: str | None = None
     ) -> QuerySet:
         """
         Returns the relationships of the given rel_type and taxonomy where:
@@ -68,7 +68,7 @@ def get_relationships(
     @classmethod
     def get_organizations(
         cls, taxonomy: Taxonomy, rel_type: RelType
-    ) -> List[Organization]:
+    ) -> list[Organization]:
         """
         Returns the list of Organizations which have the given relationship to the taxonomy.
         """
@@ -91,7 +91,7 @@ class Meta:
         proxy = True
 
     @property
-    def object_key(self) -> Union[BlockUsageLocator, LearningContextKey]:
+    def object_key(self) -> BlockUsageLocator | LearningContextKey:
         """
         Returns the object ID parsed as a UsageKey or LearningContextKey.
         Raises InvalidKeyError object_id cannot be parse into one of those key types.
@@ -115,7 +115,7 @@ class ContentTaxonomyMixin:
     def taxonomies_for_org(
         cls,
         queryset: QuerySet,
-        org: Organization = None,
+        org: Organization | None = None,
     ) -> QuerySet:
         """
         Filters the given QuerySet to those ContentTaxonomies which are available for the given organization.

openedx/features/content_tagging/models/system_defined.py

@@ -1,62 +1,14 @@
 """
 System defined models
 """
-from typing import Type
-
 from openedx_tagging.core.tagging.models import (
-    ModelSystemDefinedTaxonomy,
-    ModelObjectTag,
     UserSystemDefinedTaxonomy,
     LanguageTaxonomy,
 )
 
-from organizations.models import Organization
 from .base import ContentTaxonomyMixin
 
 
-class OrganizationModelObjectTag(ModelObjectTag):
-    """
-    ObjectTags for the OrganizationSystemDefinedTaxonomy.
-    """
-
-    class Meta:
-        proxy = True
-
-    @property
-    def tag_class_model(self) -> Type:
-        """
-        Associate the organization model
-        """
-        return Organization
-
-    @property
-    def tag_class_value(self) -> str:
-        """
-        Returns the organization name to use it on Tag.value when creating Tags for this taxonomy.
-        """
-        return "name"
-
-
-class ContentOrganizationTaxonomy(ContentTaxonomyMixin, ModelSystemDefinedTaxonomy):
-    """
-    Organization system-defined taxonomy that accepts ContentTags
-
-    Side note: The organization of an object is already encoded in its usage ID,
-    but a Taxonomy with Organization as Tags is being used so that the objects can be
-    indexed and can be filtered in the same tagging system, without any special casing.
-    """
-
-    class Meta:
-        proxy = True
-
-    @property
-    def object_tag_class(self) -> Type:
-        """
-        Returns OrganizationModelObjectTag as ObjectTag subclass associated with this taxonomy.
-        """
-        return OrganizationModelObjectTag
-
-
 class ContentLanguageTaxonomy(ContentTaxonomyMixin, LanguageTaxonomy):
     """
     Language system-defined taxonomy that accepts ContentTags