Add field disk_size to Repository & RepositoryVersion

fixes: pulp#2079

CHANGES/2079.feature

@@ -0,0 +1 @@
+Added field ``disk_size`` to Repository and RepositoryVersion that shows current total size in bytes of files stored on disk.

pulpcore/app/models/repository.py

@@ -82,6 +82,21 @@ class Meta:
         unique_together = ("name", "pulp_domain")
         verbose_name_plural = "repositories"
 
+    @property
+    def disk_size(self):
+        """Returns the approximate size on disk for all artifacts stored across all versions."""
+        all_content = (
+            RepositoryContent.objects.filter(repository=self)
+            .distinct("content")
+            .values_list("content")
+        )
+        return (
+            Artifact.objects.filter(content__pk__in=all_content)
+            .distinct()
+            .aggregate(size=models.Sum("size"))["size"]
+            or 0
+        )
+
     def on_new_version(self, version):
         """Called after a new repository version has been created.
 
@@ -795,6 +810,11 @@ def artifacts(self):
         """
         return self.repository.cast().artifacts_for_version(self)
 
+    @property
+    def disk_size(self):
+        """Returns the size on disk of all the artifacts in this repository version."""
+        return self.artifacts.distinct().aggregate(size=models.Sum("size"))["size"] or 0
+
     def added(self, base_version=None):
         """
         Args:

pulpcore/app/serializers/repository.py

@@ -4,6 +4,7 @@
 
 from rest_framework import fields, serializers
 from rest_framework_nested.serializers import NestedHyperlinkedModelSerializer
+from drf_spectacular.utils import extend_schema_field
 
 from pulpcore.app import models, settings
 from pulpcore.app.serializers import (
@@ -49,6 +50,18 @@ class RepositorySerializer(ModelSerializer):
         required=False,
         allow_null=True,
     )
+    disk_size = serializers.SerializerMethodField(
+        help_text=_(
+            "Approximate size on disk of all artifacts in the repository across all its versions."
+        ),
+    )
+
+    @extend_schema_field(serializers.IntegerField(read_only=True, allow_null=True))
+    def get_disk_size(self, obj):
+        if view := self.context.get("view", None):
+            if getattr(view, "action", "") == "retrieve":
+                return obj.disk_size
+        return None
 
     def validate_remote(self, value):
         if value and type(value) not in self.Meta.model.REMOTE_TYPES:
@@ -66,6 +79,7 @@ class Meta:
             "latest_version_href",
             "name",
             "description",
+            "disk_size",
             "retain_repo_versions",
             "remote",
         )
@@ -420,12 +434,25 @@ class RepositoryVersionSerializer(ModelSerializer, NestedHyperlinkedModelSeriali
         source="*",
         read_only=True,
     )
+    disk_size = serializers.SerializerMethodField(
+        help_text=_(
+            "Approximate size on disk of all artifacts in the repository version."
+        ),
+    )
+
+    @extend_schema_field(serializers.IntegerField(read_only=True, allow_null=True))
+    def get_disk_size(self, obj):
+        if view := self.context.get("view", None):
+            if getattr(view, "action", "") == "retrieve":
+                return obj.disk_size
+        return None
 
     class Meta:
         model = models.RepositoryVersion
         fields = ModelSerializer.Meta.fields + (
             "pulp_href",
             "number",
+            "disk_size",
             "repository",
             "base_version",
             "content_summary",

pulpcore/tests/functional/api/test_repos.py

@@ -48,3 +48,58 @@ def test_repository_content_filters(
     # but not in its latest version anymore
     results = file_repository_api_client.list(latest_with_content=content.pulp_href).results
     assert results == []
+
+
+@pytest.mark.parallel
+def test_repo_size(
+    file_repo,
+    file_repository_api_client,
+    file_repository_version_api_client,
+    file_remote_factory,
+    basic_manifest_path,
+    random_artifact_factory,
+    file_content_api_client,
+    monitor_task
+):
+    # Check that 'disk_size' is only added on retrieve operations
+    file_repo = file_repository_api_client.read(file_repo.pulp_href)
+    assert file_repo.disk_size == 0
+    repo_ver0 = file_repository_version_api_client.read(file_repo.latest_version_href)
+    assert repo_ver0.disk_size == 0
+
+    listed_repo = file_repository_api_client.list().results[0]
+    assert listed_repo.disk_size is None
+    listed_ver = file_repository_version_api_client.list(file_repo.pulp_href).results[0]
+    assert listed_ver.disk_size is None
+
+    # Sync repository with on_demand
+    remote = file_remote_factory(manifest_path=basic_manifest_path, policy="on_demand")
+    body = {"remote": remote.pulp_href}
+    monitor_task(file_repository_api_client.sync(file_repo.pulp_href, body).task)
+    file_repo = file_repository_api_client.read(file_repo.pulp_href)
+
+    # disk_size should still be 0
+    assert file_repo.disk_size == 0
+    repo_ver1 = file_repository_version_api_client.read(file_repo.latest_version_href)
+    assert repo_ver1.disk_size == 0
+
+    # Resync with immediate
+    remote = file_remote_factory(manifest_path=basic_manifest_path, policy="immediate")
+    body = {"remote": remote.pulp_href}
+    monitor_task(file_repository_api_client.sync(file_repo.pulp_href, body).task)
+    file_repo = file_repository_api_client.read(file_repo.pulp_href)
+
+    assert file_repo.disk_size == 3072  # 3 * 1024
+    repo_ver1 = file_repository_version_api_client.read(file_repo.latest_version_href)
+    assert repo_ver1.disk_size == 3072
+
+    # Add content unit w/ same name, but different artifact
+    art1 = random_artifact_factory()
+    body = {"repository": file_repo.pulp_href, "artifact": art1.pulp_href, "relative_path": "1.iso"}
+    monitor_task(file_content_api_client.create(**body).task)
+    file_repo = file_repository_api_client.read(file_repo.pulp_href)
+
+    assert file_repo.disk_size == 3072 + art1.size  # All 4 artifacts in repo
+    repo_ver2 = file_repository_version_api_client.read(file_repo.latest_version_href)
+    assert repo_ver2.content_summary.present["file.file"]["count"] == 3
+    assert repo_ver2.disk_size == 2048 + art1.size  # New size of 3 artifacts in version