[PULP-2041] Add PEP 592 yank support

CHANGES/1270.feature

@@ -0,0 +1 @@
+Added yank support (PEP 592).

docs/user/guides/package_policies.md

@@ -94,6 +94,62 @@ pulp python repository blocklist list --repository "foo"
 
 Once an entry is removed, packages matching it can be added to the repository again.
 
+## Package Yanking
+
+[PEP 592](https://peps.python.org/pep-0592/) allows marking package versions as "yanked".
+Package installers like `pip` will skip yanked versions when resolving dependencies.
+However, if a user requests an exact version (e.g. `pip install twine==5.1.0`),
+the yanked package will still be installed, with a warning.
+
+Yank status is per-repository: yanking a package in one repository does not affect other repositories
+that contain the same package.
+
+### Yank a package version
+
+To yank a package version, send a POST request to the distribution's `/yank/` endpoint:
+
+```bash
+http POST http://localhost:5001/pypi/default/<distribution-base-path>/yank/ \
+    name=shelf-reader version=0.1 yanked_reason="critical security bug" \
+    -a admin:password
+```
+
+The `yanked_reason` field is optional. If omitted, the package is marked as yanked with no reason.
+
+Yanking creates a new repository version with the yank marker added.
+Yanking a version that is already yanked with the same reason is a no-op (no new repository version is created).
+Re-yanking with a different reason will update the reason and create a new repository version.
+
+### Unyank a package version
+
+```bash
+http POST http://localhost:5001/pypi/default/<distribution-base-path>/unyank/ \
+    name=shelf-reader version=0.1 \
+    -a admin:password
+```
+
+Unyanking creates a new repository version with the yank marker removed.
+Unyanking a version that is not yanked is a no-op.
+
+### Syncing yanked packages
+
+When syncing from a remote that has yanked packages (e.g. PyPI), the yank status is preserved automatically.
+Pulp creates a yank marker for each yanked version and includes it in the repository version.
+
+### Viewing yank status
+
+Yank status is visible in the Simple API and the PyPI Metadata API.
+
+Yank markers can also be listed via the REST API:
+
+```bash
+# List all yank markers
+http GET http://localhost:5001/pulp/default/api/v3/content/python/yanks/ -a admin:password
+
+# List yank markers for a specific repository version
+http GET http://localhost:5001/pulp/default/api/v3/content/python/yanks/?repository_version=<repo-version-href> -a admin:password
+```
+
 ## Package Substitution
 
 By default, Python repositories allow package substitution: uploading, syncing, or adding a package

pulp_python/app/migrations/0023_packageyank.py

@@ -0,0 +1,50 @@
+# Generated by Django 5.2.16 on 2026-07-30 10:44
+
+import django.db.models.deletion
+from django.db import migrations, models
+
+import pulpcore.app.util
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("python", "0022_pythonblocklistentry"),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name="PackageYank",
+            fields=[
+                (
+                    "content_ptr",
+                    models.OneToOneField(
+                        auto_created=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        parent_link=True,
+                        primary_key=True,
+                        serialize=False,
+                        to="core.content",
+                    ),
+                ),
+                ("name_normalized", models.TextField()),
+                ("version", models.TextField()),
+                ("yanked_reason", models.TextField(default="")),
+                (
+                    "_pulp_domain",
+                    models.ForeignKey(
+                        default=pulpcore.app.util.get_domain_pk,
+                        on_delete=django.db.models.deletion.PROTECT,
+                        to="core.domain",
+                    ),
+                ),
+            ],
+            options={
+                "default_related_name": "%(app_label)s_%(model_name)s",
+                "unique_together": {
+                    ("name_normalized", "version", "yanked_reason", "_pulp_domain")
+                },
+            },
+            bases=("core.content",),
+        ),
+    ]

pulp_python/app/models.py

@@ -204,8 +204,6 @@ class PythonPackageContent(Content):
     sha256 = models.CharField(db_index=True, max_length=64)
     metadata_sha256 = models.CharField(max_length=64, null=True)
     size = models.BigIntegerField(default=0)
-    # yanked and yanked_reason are not implemented because they are mutable
-
     # From pulpcore
     PROTECTED_FROM_RECLAIM = False
     TYPE = "python"
@@ -289,6 +287,32 @@ class Meta:
         unique_together = ("sha256", "_pulp_domain")
 
 
+class PackageYank(Content):
+    """
+    A marker content type indicating a package version is yanked in a repository (PEP 592).
+
+    Its presence in a repository version means all files for the matching
+    (name_normalized, version) pair are yanked. Yank/unyank operations
+    add/remove this marker, creating new repository versions.
+    """
+
+    TYPE = "python_yank"
+    repo_key_fields = ("name_normalized", "version")
+
+    name_normalized = models.TextField()
+    version = models.TextField()
+    yanked_reason = models.TextField(default="")
+
+    _pulp_domain = models.ForeignKey("core.Domain", default=get_domain_pk, on_delete=models.PROTECT)
+
+    def __str__(self):
+        return f"<{self._meta.object_name}: {self.name_normalized} [{self.version}]>"
+
+    class Meta:
+        default_related_name = "%(app_label)s_%(model_name)s"
+        unique_together = ("name_normalized", "version", "yanked_reason", "_pulp_domain")
+
+
 class PythonPublication(Publication, AutoAddObjPermsMixin):
     """
     A Publication for PythonContent.
@@ -364,7 +388,7 @@ class PythonRepository(Repository, AutoAddObjPermsMixin):
     """
 
     TYPE = "python"
-    CONTENT_TYPES = [PythonPackageContent, PackageProvenance]
+    CONTENT_TYPES = [PythonPackageContent, PackageProvenance, PackageYank]
     REMOTE_TYPES = [PythonRemote]
     PULL_THROUGH_SUPPORTED = True
 

pulp_python/app/pypi/serializers.py

@@ -136,3 +136,24 @@ class PackageUploadTaskSerializer(serializers.Serializer):
     session = serializers.CharField(allow_null=True)
     task = serializers.CharField()
     task_start_time = serializers.DateTimeField(allow_null=True)
+
+
+class YankSerializer(serializers.Serializer):
+    """
+    A Serializer for yank/unyank requests (PEP 592).
+    """
+
+    name = serializers.CharField(
+        help_text=_("The name of the package to yank or unyank."),
+        required=True,
+    )
+    version = serializers.CharField(
+        help_text=_("The version of the package to yank or unyank."),
+        required=True,
+    )
+    yanked_reason = serializers.CharField(
+        help_text=_("The reason for yanking the package version."),
+        required=False,
+        allow_blank=True,
+        default="",
+    )

pulp_python/app/pypi/views.py

@@ -37,6 +37,7 @@
 from pulp_python.app.cache import PythonApiCache, find_base_path_cached
 from pulp_python.app.models import (
     PackageProvenance,
+    PackageYank,
     PythonDistribution,
     PythonPackageContent,
     PythonPublication,
@@ -46,6 +47,7 @@
     PackageUploadSerializer,
     PackageUploadTaskSerializer,
     SummarySerializer,
+    YankSerializer,
 )
 from pulp_python.app.utils import (
     PYPI_LAST_SERIAL,
@@ -356,6 +358,8 @@ def parse_package(release_package):
                 "upload_time": release_package.upload_time,
                 "version": release_package.version,
                 "provenance": release_package.provenance_url,
+                "yanked": release_package.is_yanked,
+                "yanked_reason": release_package.yanked_reason or "",
             }
 
         rfilter = get_remote_package_filter(remote)
@@ -408,6 +412,11 @@ def retrieve(self, request, path, package):
                 "version",
                 "has_provenance",
             )
+            yank_markers = dict(
+                PackageYank.objects.filter(
+                    pk__in=repo_ver.content, name_normalized=normalized
+                ).values_list("version", "yanked_reason")
+            )
             local_releases = {
                 p["filename"]: {
                     **p,
@@ -418,6 +427,8 @@ def retrieve(self, request, path, package):
                         if p["has_provenance"]
                         else None
                     ),
+                    "yanked": p["version"] in yank_markers,
+                    "yanked_reason": yank_markers.get(p["version"], ""),
                 }
                 for p in packages
             }
@@ -493,12 +504,18 @@ def retrieve(self, request, path, meta):
             headers = {PYPI_LAST_SERIAL: str(PYPI_SERIAL_CONSTANT)}
             if settings.DOMAIN_ENABLED:
                 domain = get_domain()
+            yank_markers = dict(
+                PackageYank.objects.filter(
+                    pk__in=repo_ver.content, name_normalized=normalized
+                ).values_list("version", "yanked_reason")
+            )
             json_body = python_content_to_json(
                 path,
                 package_content,
                 version=version,
                 domain=domain,
                 repository_version=repo_ver,
+                yank_markers=yank_markers,
             )
             if json_body:
                 return Response(data=json_body, headers=headers)
@@ -586,3 +603,70 @@ def retrieve(self, request, path, package, version, filename):
                 if provenance:
                     return Response(data=provenance.provenance)
         return HttpResponseNotFound(f"{package} {version} {filename} provenance does not exist.")
+
+
+class YankView(PyPIMixin, ViewSet):
+    """View for yank/unyank requests (PEP 592)."""
+
+    endpoint_name = "yank"
+    DEFAULT_ACCESS_POLICY = {
+        "statements": [
+            {
+                "action": ["yank", "unyank"],
+                "principal": "authenticated",
+                "effect": "allow",
+                "condition": "index_has_repo_perm:python.modify_pythonrepository",
+            },
+        ],
+    }
+
+    @extend_schema(request=YankSerializer, summary="Yank a package version")
+    def yank(self, request, path):
+        """Yank a package version, marking all its files with data-yanked."""
+        repo = self.distribution.repository
+        if not repo:
+            return HttpResponseBadRequest(reason="Index is not pointing to a repository")
+
+        serializer = YankSerializer(data=request.data)
+        serializer.is_valid(raise_exception=True)
+
+        normalized = canonicalize_name(serializer.validated_data["name"])
+        version = serializer.validated_data["version"]
+        repo_ver = self.get_repository_version(self.distribution)
+        if not PythonPackageContent.objects.filter(
+            pk__in=repo_ver.content, name_normalized=normalized, version=version
+        ).exists():
+            return HttpResponseNotFound(f"{normalized}=={version} not found in repository")
+
+        result = dispatch(
+            tasks.ayank_package,
+            exclusive_resources=[repo],
+            kwargs={
+                "repository_pk": str(repo.pk),
+                "name": serializer.validated_data["name"],
+                "version": serializer.validated_data["version"],
+                "yanked_reason": serializer.validated_data.get("yanked_reason", ""),
+            },
+        )
+        return OperationPostponedResponse(result, request)
+
+    @extend_schema(request=YankSerializer, summary="Unyank a package version")
+    def unyank(self, request, path):
+        """Unyank a package version, unmarking all its files with data-yanked."""
+        repo = self.distribution.repository
+        if not repo:
+            return HttpResponseBadRequest(reason="Index is not pointing to a repository")
+
+        serializer = YankSerializer(data=request.data)
+        serializer.is_valid(raise_exception=True)
+
+        result = dispatch(
+            tasks.aunyank_package,
+            exclusive_resources=[repo],
+            kwargs={
+                "repository_pk": str(repo.pk),
+                "name": serializer.validated_data["name"],
+                "version": serializer.validated_data["version"],
+            },
+        )
+        return OperationPostponedResponse(result, request)

pulp_python/app/serializers.py

@@ -683,6 +683,25 @@ class Meta:
         model = python_models.PackageProvenance
 
 
+class PackageYankSerializer(core_serializers.NoArtifactContentSerializer):
+    """
+    Read-only serializer for PackageYank content units (PEP 592).
+    Used by PackageYankViewSet to expose yank markers via the Pulp REST API.
+    """
+
+    name_normalized = serializers.CharField(read_only=True)
+    version = serializers.CharField(read_only=True)
+    yanked_reason = serializers.CharField(read_only=True)
+
+    class Meta:
+        fields = core_serializers.NoArtifactContentSerializer.Meta.fields + (
+            "name_normalized",
+            "version",
+            "yanked_reason",
+        )
+        model = python_models.PackageYank
+
+
 class MultipleChoiceArrayField(serializers.MultipleChoiceField):
     """
     A wrapper to make sure this DRF serializer works properly with ArrayFields.

pulp_python/app/tasks/__init__.py

@@ -7,3 +7,4 @@
 from .sync import sync  # noqa:F401
 from .upload import upload, upload_group  # noqa:F401
 from .vulnerability_report import get_repo_version_content  # noqa:F401
+from .yank import aunyank_package, ayank_package  # noqa:F401

pulp_python/app/tasks/sync.py

@@ -24,6 +24,7 @@
 from pulp_python.app.exceptions import UnsupportedProtocolError
 from pulp_python.app.models import (
     PackageProvenance,
+    PackageYank,
     PythonPackageContent,
     PythonRemote,
 )
@@ -265,6 +266,15 @@ async def create_content(self, pkg):
                         )
                         d_artifacts.append(metadata_artifact)
 
+                    if upstream_pkg.is_yanked:
+                        yank_marker = PackageYank(
+                            name_normalized=pkg.name,
+                            version=version,
+                            yanked_reason=upstream_pkg.yanked_reason or "",
+                        )
+                        yank_dc = DeclarativeContent(content=yank_marker, d_artifacts=[])
+                        await self.python_stage.put(yank_dc)
+
                 dc = DeclarativeContent(content=package, d_artifacts=d_artifacts)
                 declared_contents[entry["filename"]] = dc
                 await self.python_stage.put(dc)