Add Simple API integration

jobselko · jobselko · commit 5db692efde1c · 2026-07-24T17:03:05.000+02:00
diff --git a/pulp_python/app/pypi/views.py b/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,
@@ -384,6 +385,7 @@ def retrieve(self, request, path, package):
         normalized = canonicalize_name(package)
         releases = {}
         if self.distribution.remote:
+            # todo: what if no repo?
             releases = self.pull_through_package_simple(normalized, path, self.distribution.remote)
         elif self.should_redirect(repo_version=repo_ver):
             return redirect(urljoin(self.base_content_url, f"{path}/simple/{normalized}/"))
@@ -409,6 +411,11 @@ 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,
@@ -419,6 +426,8 @@ 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
             }
diff --git a/pulp_python/app/tasks/sync.py b/pulp_python/app/tasks/sync.py
@@ -25,6 +25,7 @@
 from pulp_python.app.models import (
     PackageProvenance,
     PythonPackageContent,
+    PackageYank,
     PythonRemote,
 )
 from pulp_python.app.provenance import Provenance
@@ -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)
diff --git a/pulp_python/app/utils.py b/pulp_python/app/utils.py
@@ -50,7 +50,6 @@
 </html>
 """
 
-# TODO in the future: data-yanked (not implemented yet because it is mutable)
 simple_detail_template = """<!DOCTYPE html>
 <html>
   <head>
@@ -62,6 +61,7 @@
     {%- for pkg in project_packages %}
       <a href="{{ pkg.url }}#sha256={{ pkg.sha256 }}" rel="internal"
       {%- if pkg.requires_python %} data-requires-python="{{ pkg.requires_python }}" {%- endif %}
+      {%- if pkg.yanked %} data-yanked="{{ pkg.yanked_reason }}" {%- endif %}
       {%- if pkg.metadata_sha256 %} data-dist-info-metadata="sha256={{ pkg.metadata_sha256 }}" data-core-metadata="sha256={{ pkg.metadata_sha256 }}"
       {%- endif %} {% if pkg.provenance -%}
       data-provenance="{{ pkg.provenance }}"{% endif %}>{{ pkg.filename }}</a><br/>
@@ -591,7 +591,12 @@ def write_simple_detail_json(project_name, project_packages):
                 "core-metadata": (
                     {"sha256": package["metadata_sha256"]} if package["metadata_sha256"] else False
                 ),
-                # yanked and yanked_reason are not implemented because they are mutable
+                # PEP 592
+                "yanked": (
+                    package["yanked_reason"]
+                    if package["yanked"] and package["yanked_reason"]
+                    else package["yanked"]
+                ),
                 # (v1.1, PEP 700)
                 "size": package["size"],
                 "upload-time": format_upload_time(package["upload_time"]),
diff --git a/pulp_python/tests/functional/api/test_yank.py b/pulp_python/tests/functional/api/test_yank.py
@@ -4,6 +4,7 @@
 import requests
 
 from pulp_python.tests.functional.constants import (
+    PYPI_SIMPLE_V1_JSON,
     PYTHON_FIXTURES_URL,
     TWINE_EGG_FILENAME,
     TWINE_EGG_URL,
@@ -66,6 +67,16 @@ def test_yank_and_unyank(
     summary = python_content_summary(repository=repo, version=2)
     assert summary.added["python.python_yank"]["count"] == 1
 
+    # Simple API JSON check - yanked files should have yanked reason
+    simple_url = f"{urljoin(distro.base_url, 'simple/')}{TWINE_NAME}"
+    response = requests.get(simple_url, headers={"Accept": PYPI_SIMPLE_V1_JSON})
+    assert response.json()["files"][0]["yanked"] == "broken"
+    assert response.json()["files"][1]["yanked"] == "broken"
+
+    # Simple API HTML check - data-yanked attribute should be present
+    response = requests.get(simple_url)
+    assert response.text.count('data-yanked="broken"') == 2
+
     # Yank again - idempotent, no new repo version
     response = yank(distro, TWINE_NAME, TWINE_VERSION, yanked_reason="broken")
     assert response.status_code == 202
@@ -85,6 +96,15 @@ def test_yank_and_unyank(
     summary = python_content_summary(repository=repo, version=3)
     assert summary.removed["python.python_yank"]["count"] == 1
 
+    # Simple API JSON - yanked should be False after unyank
+    response = requests.get(simple_url, headers={"Accept": PYPI_SIMPLE_V1_JSON})
+    assert response.json()["files"][0]["yanked"] == False
+    assert response.json()["files"][1]["yanked"] == False
+
+    # Simple API HTML - data-yanked should not be present
+    response = requests.get(simple_url)
+    assert "data-yanked" not in response.text
+
     # Unyank again - idempotent, no new repo version
     response = unyank(distro, TWINE_NAME, TWINE_VERSION)
     assert response.status_code == 202
@@ -94,6 +114,36 @@ def test_yank_and_unyank(
     assert repo.latest_version_href == version_3
 
 
+def test_yank_sync(
+    delete_orphans_pre,
+    python_remote_factory,
+    python_repo_with_sync,
+    python_content_summary,
+    python_distribution_factory,
+):
+    """
+    Syncing a yanked package from upstream creates a PackageYank marker.
+    """
+    remote = python_remote_factory(includes=["twine==5.1.0"])
+    repo = python_repo_with_sync(remote)
+    distro = python_distribution_factory(repository=repo)
+
+    # Sync should have created a yank marker for twine 5.1.0
+    summary = python_content_summary(repository=repo, version=1)
+    assert summary.added["python.python"]["count"] == 2
+    assert summary.added["python.python_yank"]["count"] == 1
+
+    # Simple API JSON check - yanked reason from upstream
+    simple_url = f"{urljoin(distro.base_url, 'simple/')}{TWINE_NAME}"
+    response = requests.get(simple_url, headers={"Accept": PYPI_SIMPLE_V1_JSON})
+    assert response.json()["files"][0]["yanked"] == "https://github.com/pypa/twine/issues/1125"
+    assert response.json()["files"][1]["yanked"] == "https://github.com/pypa/twine/issues/1125"
+
+    # Simple API HTML check - data-yanked attribute present
+    response = requests.get(simple_url)
+    assert response.text.count("data-yanked=") == 2
+
+
 @pytest.mark.parallel
 def test_partial_yank(
     monitor_task,
