mature -> sensitive code changes for the API (#3769)

* Make database table name explicit for "mature" tables

* Rename mature to sensitive

* Use the sensitive source field in the serializer

Also correctly reference mature vs sensitive depending on what type of object is being handed into the serializer

api/api/admin/__init__.py

@@ -2,7 +2,7 @@
 
 from api.admin.site import openverse_admin
 from api.models import PENDING, Audio, AudioReport, ContentProvider, Image, ImageReport
-from api.models.media import AbstractDeletedMedia, AbstractMatureMedia
+from api.models.media import AbstractDeletedMedia, AbstractSensitiveMedia
 
 
 admin.site = openverse_admin
@@ -72,7 +72,7 @@ def has_add_permission(self, *args, **kwargs):
 
 
 for klass in [
-    *AbstractMatureMedia.__subclasses__(),
+    *AbstractSensitiveMedia.__subclasses__(),
     *AbstractDeletedMedia.__subclasses__(),
 ]:
     admin.site.register(klass, MediaSubreportAdmin)

api/api/migrations/0055_alter_matureaudio_table_alter_matureimage_table.py

@@ -0,0 +1,21 @@
+# Generated by Django 4.2.9 on 2024-02-08 01:52
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('api', '0054_throttledapplication_post_logout_redirect_uris'),
+    ]
+
+    operations = [
+        migrations.AlterModelTable(
+            name='matureaudio',
+            table='api_matureaudio',
+        ),
+        migrations.AlterModelTable(
+            name='matureimage',
+            table='api_matureimage',
+        ),
+    ]

api/api/migrations/0056_rename_mature_to_sensitive.py

@@ -0,0 +1,27 @@
+# Generated by Django 4.2.9 on 2024-02-08 18:26
+
+import api.models.media
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('api', '0055_alter_matureaudio_table_alter_matureimage_table'),
+    ]
+
+    operations = [
+        migrations.RenameModel('MatureImage', 'SensitiveImage'),
+        migrations.RenameModel('MatureAudio', 'SensitiveAudio'),
+        migrations.AlterField(
+            model_name='sensitiveaudio',
+            name='media_obj',
+            field=models.OneToOneField(db_column='identifier', db_constraint=False, help_text='The reference to the sensitive audio.', on_delete=django.db.models.deletion.DO_NOTHING, primary_key=True, related_name='sensitive_audio', serialize=False, to='api.audio', to_field='identifier'),
+        ),
+        migrations.AlterField(
+            model_name='sensitiveimage',
+            name='media_obj',
+            field=models.OneToOneField(db_column='identifier', db_constraint=False, help_text='The reference to the sensitive image.', on_delete=django.db.models.deletion.DO_NOTHING, primary_key=True, related_name='sensitive_image', serialize=False, to='api.image', to_field='identifier'),
+        ),
+    ]

api/api/models/__init__.py

@@ -6,9 +6,9 @@
     AudioReport,
     AudioSet,
     DeletedAudio,
-    MatureAudio,
+    SensitiveAudio,
 )
-from api.models.image import DeletedImage, Image, ImageList, ImageReport, MatureImage
+from api.models.image import DeletedImage, Image, ImageList, ImageReport, SensitiveImage
 from api.models.media import (
     DEINDEXED,
     DMCA,

api/api/models/audio.py

@@ -9,10 +9,10 @@
 from api.models.media import (
     AbstractAltFile,
     AbstractDeletedMedia,
-    AbstractMatureMedia,
     AbstractMedia,
     AbstractMediaList,
     AbstractMediaReport,
+    AbstractSensitiveMedia,
 )
 from api.models.mixins import FileMixin, ForeignIdentifierMixin, MediaMixin
 from api.utils.waveform import generate_peaks
@@ -189,8 +189,8 @@ class Audio(AudioFileMixin, AbstractMedia):
     )
 
     @property
-    def mature(self) -> bool:
-        return hasattr(self, "mature_audio")
+    def sensitive(self) -> bool:
+        return hasattr(self, "sensitive_audio")
 
     @property
     def alternative_files(self):
@@ -260,7 +260,7 @@ class Meta:
         verbose_name_plural = "Deleted audio"
 
 
-class MatureAudio(AbstractMatureMedia):
+class SensitiveAudio(AbstractSensitiveMedia):
     """
     Stores all audio tracks that have been flagged as 'mature'.
 
@@ -278,17 +278,18 @@ class MatureAudio(AbstractMatureMedia):
         primary_key=True,
         db_constraint=False,
         db_column="identifier",
-        related_name="mature_audio",
+        related_name="sensitive_audio",
         help_text="The reference to the sensitive audio.",
     )
 
     class Meta:
+        db_table = "api_matureaudio"
         verbose_name_plural = "Mature audio"
 
 
 class AudioReport(AbstractMediaReport):
     media_class = Audio
-    mature_class = MatureAudio
+    sensitive_class = SensitiveAudio
     deleted_class = DeletedAudio
 
     media_obj = models.ForeignKey(

api/api/models/image.py

@@ -6,10 +6,10 @@
 from api.constants.media_types import IMAGE_TYPE
 from api.models.media import (
     AbstractDeletedMedia,
-    AbstractMatureMedia,
     AbstractMedia,
     AbstractMediaList,
     AbstractMediaReport,
+    AbstractSensitiveMedia,
 )
 from api.models.mixins import FileMixin
 
@@ -53,8 +53,8 @@ class Meta(AbstractMedia.Meta):
         db_table = "image"
 
     @property
-    def mature(self) -> bool:
-        return hasattr(self, "mature_image")
+    def sensitive(self) -> bool:
+        return hasattr(self, "sensitive_image")
 
 
 class DeletedImage(AbstractDeletedMedia):
@@ -80,7 +80,7 @@ class DeletedImage(AbstractDeletedMedia):
     )
 
 
-class MatureImage(AbstractMatureMedia):
+class SensitiveImage(AbstractSensitiveMedia):
     """
     Stores all images that have been flagged as 'mature'.
 
@@ -98,14 +98,17 @@ class MatureImage(AbstractMatureMedia):
         primary_key=True,
         db_constraint=False,
         db_column="identifier",
-        related_name="mature_image",
+        related_name="sensitive_image",
         help_text="The reference to the sensitive image.",
     )
 
+    class Meta:
+        db_table = "api_matureimage"
+
 
 class ImageReport(AbstractMediaReport):
     media_class = Image
-    mature_class = MatureImage
+    sensitive_class = SensitiveImage
     deleted_class = DeletedImage
 
     media_obj = models.ForeignKey(

api/api/models/media.py

@@ -133,15 +133,15 @@ class AbstractMediaReport(models.Model):
     """
     Generic model from which to inherit all reported media classes.
 
-    'Reported' here refers to content reports such as mature, copyright-violating or
-    deleted content. Subclasses must populate ``media_class``, ``mature_class`` and
+    'Reported' here refers to content reports such as sensitive, copyright-violating or
+    deleted content. Subclasses must populate ``media_class``, ``sensitive_class`` and
     ``deleted_class`` fields.
     """
 
     media_class: type[models.Model] = None
     """the model class associated with this media type e.g. ``Image`` or ``Audio``"""
-    mature_class: type[models.Model] = None
-    """the class storing mature media e.g. ``MatureImage`` or ``MatureAudio``"""
+    sensitive_class: type[models.Model] = None
+    """the class storing sensitive media e.g. ``SensitiveImage`` or ``SensitiveAudio``"""
     deleted_class: type[models.Model] = None
     """the class storing deleted media e.g. ``DeletedImage`` or ``DeletedAudio``"""
 
@@ -213,18 +213,18 @@ def save(self, *args, **kwargs):
         Extend the built-in ``save()`` functionality of Django with Elasticsearch
         integration to update records and refresh indices.
 
-        Media marked as mature or deleted also leads to instantiation of their
-        corresponding mature or deleted classes.
+        Media marked as sensitive or deleted also leads to instantiation of their
+        corresponding sensitive or deleted classes.
         """
 
         self.clean()
 
         super().save(*args, **kwargs)
 
         if self.status == MATURE_FILTERED:
-            # Create an instance of the mature class for this media. This will
+            # Create an instance of the sensitive class for this media. This will
             # automatically set the ``mature`` field in the ES document.
-            self.mature_class.objects.create(media_obj=self.media_obj)
+            self.sensitive_class.objects.create(media_obj=self.media_obj)
         elif self.status == DEINDEXED:
             # Create an instance of the deleted class for this media, so that we don't
             # reindex it later. This will automatically delete the ES document and the
@@ -290,7 +290,7 @@ class AbstractDeletedMedia(PerformIndexUpdateMixin, OpenLedgerModel):
     Generic model from which to inherit all deleted media classes.
 
     'Deleted' here refers to media which has been deleted at the source or intentionally
-    de-indexed by us. Unlike mature reports, this action is irreversible. Subclasses
+    de-indexed by us. Unlike sensitive reports, this action is irreversible. Subclasses
     must populate ``media_class`` and ``es_index`` fields.
     """
 
@@ -329,9 +329,9 @@ def save(self, *args, **kwargs):
         self.media_obj.delete()  # remove the actual model instance
 
 
-class AbstractMatureMedia(PerformIndexUpdateMixin, models.Model):
+class AbstractSensitiveMedia(PerformIndexUpdateMixin, models.Model):
     """
-    Generic model from which to inherit all mature media classes.
+    Generic model from which to inherit all sensitive media classes.
 
     Subclasses must populate ``media_class`` and ``es_index`` fields.
     """
@@ -350,7 +350,7 @@ class AbstractMatureMedia(PerformIndexUpdateMixin, models.Model):
         primary_key=True,
         db_constraint=False,
         db_column="identifier",
-        related_name="mature_abstract_media",
+        related_name="sensitive_abstract_media",
         help_text="The reference to the sensitive media.",
     )
     """

api/api/serializers/media_serializers.py

@@ -516,6 +516,7 @@ class Meta:
 
     mature = serializers.BooleanField(
         help_text="Whether the media item is marked as mature",
+        source="sensitive",
     )
 
     # This should be promoted to a stable field alongside
@@ -540,9 +541,9 @@ def get_unstable__sensitivity(self, obj: Hit | AbstractMedia) -> list[str]:
         ):
             result.append(sensitivity.TEXT)
 
-        # ``obj.mature`` will either be `mature` from the ES document
-        # or the ``mature`` property on the Image or Audio model.
-        if obj.mature:
+        # ``obj.sensitive`` will either be `mature` from the ES document (see below)
+        # or the ``sensitive`` property on the Image or Audio model.
+        if obj.sensitive:
             # We do not currently have any documents marked `mature=true`
             # that were not marked so as a result of a confirmed user report.
             # This is despite the fact that the ingestion server _does_ copy
@@ -569,6 +570,16 @@ def get_unstable__sensitivity(self, obj: Hit | AbstractMedia) -> list[str]:
         return result
 
     def to_representation(self, *args, **kwargs):
+        # This serializer adapts both ES Hits *and* Media instances. Currently,
+        # ES has a `mature` field on it which represents if maturity was present on
+        # the record in the database. The attributes in the code have been renamed
+        # to `sensitive`, but for the time being this flag still exists on the ES index.
+        # In order to prevent failures in serialization (since the serializer is looking
+        # for the `sensitive` attribute), we rename it here.
+        obj = args[0]
+        if isinstance(obj, Hit):
+            obj.sensitive = obj.mature
+
         output = super().to_representation(*args, **kwargs)
 
         # Ensure lists are ``[]`` instead of ``None``

api/api/views/audio_views.py

@@ -50,7 +50,7 @@ class AudioViewSet(MediaViewSet):
     collection_serializer_class = AudioCollectionRequestSerializer
 
     def get_queryset(self):
-        return super().get_queryset().select_related("mature_audio", "audioset")
+        return super().get_queryset().select_related("sensitive_audio", "audioset")
 
     # Extra actions
     @creator_collection

api/api/views/image_views.py

@@ -63,7 +63,7 @@ class ImageViewSet(MediaViewSet):
     }
 
     def get_queryset(self):
-        return super().get_queryset().select_related("mature_image")
+        return super().get_queryset().select_related("sensitive_image")
 
     # Extra actions
     @creator_collection

api/test/factory/models/__init__.py

@@ -2,12 +2,12 @@
     AudioAddOnFactory,
     AudioFactory,
     AudioReportFactory,
-    MatureAudioFactory,
+    SensitiveAudioFactory,
 )
 from test.factory.models.image import (
     ImageFactory,
     ImageReportFactory,
-    MatureImageFactory,
+    SensitiveImageFactory,
 )
 from test.factory.models.oauth2 import (
     AccessTokenFactory,

api/test/factory/models/audio.py

@@ -1,20 +1,20 @@
 import factory
 from factory.django import DjangoModelFactory
 
-from api.models.audio import Audio, AudioAddOn, AudioReport, MatureAudio
+from api.models.audio import Audio, AudioAddOn, AudioReport, SensitiveAudio
 from test.factory.faker import Faker
 from test.factory.models.media import IdentifierFactory, MediaFactory
 
 
-class MatureAudioFactory(DjangoModelFactory):
+class SensitiveAudioFactory(DjangoModelFactory):
     class Meta:
-        model = MatureAudio
+        model = SensitiveAudio
 
     media_obj = factory.SubFactory("test.factory.models.audio.AudioFactory")
 
 
 class AudioFactory(MediaFactory):
-    _mature_factory = MatureAudioFactory
+    _sensitive_factory = SensitiveAudioFactory
 
     class Meta:
         model = Audio

api/test/factory/models/image.py

@@ -1,19 +1,19 @@
 import factory
 from factory.django import DjangoModelFactory
 
-from api.models.image import Image, ImageReport, MatureImage
+from api.models.image import Image, ImageReport, SensitiveImage
 from test.factory.models.media import MediaFactory, MediaReportFactory
 
 
-class MatureImageFactory(DjangoModelFactory):
+class SensitiveImageFactory(DjangoModelFactory):
     class Meta:
-        model = MatureImage
+        model = SensitiveImage
 
     media_obj = factory.SubFactory("test.factory.models.image.ImageFactory")
 
 
 class ImageFactory(MediaFactory):
-    _mature_factory = MatureImageFactory
+    _sensitive_factory = SensitiveImageFactory
 
     class Meta:
         model = Image

api/test/factory/models/media.py

@@ -37,8 +37,8 @@ class MediaFactory(DjangoModelFactory):
     )
 
     # Sub-factories must set this to their corresponding
-    # ``AbstractMatureMedia`` subclass
-    _mature_factory = None
+    # ``AbstractSensitiveMedia`` subclass
+    _sensitive_factory = None
 
     _highest_pre_existing_pk = None
 
@@ -126,7 +126,7 @@ def create(cls, *args, **kwargs) -> AbstractMedia | tuple[AbstractMedia, Hit]:
             hit = None
 
         if mature_reported:
-            cls._mature_factory.create(media_obj=model)
+            cls._sensitive_factory.create(media_obj=model)
 
         if pook_active:
             # Reactivate pook if it was active