implement unnesting and add a temporary changelog

Author: Andrew Cheng

CHANGES/+base-version-content-filter.feature

@@ -0,0 +1,6 @@
+Added a ``base_version`` filter to the content list endpoints. When combined with
+``repository_version_added`` or ``repository_version_removed``, it returns the net set of content
+added or removed between two arbitrary repository versions instead of only the single-step
+difference against the filtered version's immediate predecessor. The diff is computed database-side
+via ``unnest`` subqueries over each version's ``content_ids``, avoiding loading the arrays into
+memory or passing them as per-query parameters.

pulpcore/app/models/repository.py

@@ -1000,13 +1000,26 @@ def get_content(self, content_qs=None):
         content_ids = self.content_ids
         if len(content_ids) >= 65535:
             # Workaround for PostgreSQL's limit on the number of parameters in a query
-            content_ids = (
-                RepositoryVersion.objects.filter(pk=self.pk)
-                .annotate(cids=Func(F("content_ids"), function="unnest"))
-                .values_list("cids", flat=True)
-            )
+            content_ids = self.content_ids_subquery()
         return content_qs.filter(pk__in=content_ids)
 
+    def content_ids_subquery(self):
+        """
+        Return this version's ``content_ids`` as a database-side ``unnest`` subquery.
+
+        Using a subquery keeps the content unit UUIDs inside PostgreSQL instead of loading the
+        whole array into Python and passing each UUID as a bound query parameter. This avoids the
+        per-query parameter limit and the memory/serialization cost for large repository versions.
+
+        Returns:
+            django.db.models.QuerySet: A values queryset yielding the content unit UUIDs.
+        """
+        return (
+            RepositoryVersion.objects.filter(pk=self.pk)
+            .annotate(cids=Func(F("content_ids"), function="unnest"))
+            .values_list("cids", flat=True)
+        )
+
     @property
     def content(self):
         """
@@ -1119,8 +1132,8 @@ def added(self, base_version=None):
         if not base_version:
             return Content.objects.filter(version_memberships__version_added=self)
 
-        return Content.objects.filter(pk__in=self.content_ids).exclude(
-            pk__in=base_version.content_ids
+        return Content.objects.filter(pk__in=self.content_ids_subquery()).exclude(
+            pk__in=base_version.content_ids_subquery()
         )
 
     def removed(self, base_version=None):
@@ -1134,8 +1147,8 @@ def removed(self, base_version=None):
         if not base_version:
             return Content.objects.filter(version_memberships__version_removed=self)
 
-        return Content.objects.filter(pk__in=base_version.content_ids).exclude(
-            pk__in=self.content_ids
+        return Content.objects.filter(pk__in=base_version.content_ids_subquery()).exclude(
+            pk__in=self.content_ids_subquery()
         )
 
     def contains(self, content):