From e342d14e2952b6fedc4232fb06ae76f2a03aaf38 Mon Sep 17 00:00:00 2001 From: Matthias Dellweg Date: Thu, 16 Jul 2026 16:41:38 +0200 Subject: [PATCH] Improve content_ids handling by defered loading The array_field can be huge and most of the time we don't need the values in python. Some uses can even be handled by clever SQL without ever channelling all the data through the network. --- pulpcore/app/models/repository.py | 59 ++++++++++++++++++++++--------- 1 file changed, 42 insertions(+), 17 deletions(-) diff --git a/pulpcore/app/models/repository.py b/pulpcore/app/models/repository.py index 5be034f5ff..782c49ac53 100644 --- a/pulpcore/app/models/repository.py +++ b/pulpcore/app/models/repository.py @@ -908,6 +908,37 @@ def with_content(self, content): return self.filter(content_ids__overlap=content_pks) +class RepositoryVersionManager(models.Manager): + # `RepositoryVersionQuerySet.as_manager()` does not allow us to redefine `get_queryset`. + # Sadly, we have to replicate the filtering amenities too. + # + def get_queryset(self): + # Prevent the content_ids to be automatically hydrated. + return RepositoryVersionQuerySet(self.model, using=self._db).defer("content_ids") + + def complete(self): + return self.get_queryset().filter(complete=True) + + def with_content(self, content): + """ + Filters repository versions that contain the provided content units. + + Args: + content (django.db.models.QuerySet or list): Content queryset or list of PKs + + Returns: + django.db.models.QuerySet: Repository versions which contains content. + """ + if isinstance(content, models.QuerySet): + content_pks = content.values_list("pk", flat=True) + elif not content: + return self.none() + else: + content_pks = content + + return self.get_queryset().filter(content_ids__overlap=content_pks) + + class RepositoryVersion(BaseModel): """ A version of a repository's content set. @@ -936,7 +967,7 @@ class RepositoryVersion(BaseModel): base_version (models.ForeignKey): The repository version this was created from. """ - objects = RepositoryVersionQuerySet.as_manager() + objects = RepositoryVersionManager() repository = models.ForeignKey(Repository, on_delete=models.CASCADE) number = models.PositiveIntegerField(db_index=True) @@ -997,15 +1028,13 @@ def get_content(self, content_qs=None): if content_qs is None: content_qs = Content.objects - 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) - ) - return content_qs.filter(pk__in=content_ids) + # Try to not even attempt to evaluate the content_ids on the python side. + content_ids_subquery = ( + RepositoryVersion.objects.filter(pk=self.pk) + .annotate(cids=Func(F("content_ids"), function="unnest")) + .values_list("cids", flat=True) + ) + return content_qs.filter(pk__in=content_ids_subquery) @property def content(self): @@ -1119,9 +1148,7 @@ 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).exclude(pk__in=base_version.content) def removed(self, base_version=None): """ @@ -1134,9 +1161,7 @@ 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).exclude(pk__in=self.content) def contains(self, content): """ @@ -1145,7 +1170,7 @@ def contains(self, content): Returns: bool: True if the repository version contains the content, False otherwise """ - return content.pk in self.content_ids + return content.pk in self.content def add_content(self, content): """