Fix pull-through metadata serving

fixes: #1083

Author: gerrod3

CHANGES/1083.bugfix

@@ -0,0 +1 @@
+Fixed pull-through PEP 658 metadata not being served correctly for certain tools.

docs/user/guides/publish.md

@@ -105,6 +105,10 @@ pulp python distribution update --name foo --remote bar
     Functionality may not work or may be incomplete. Also, backwards compatibility when upgrading
     is not guaranteed.
 
+!!! warning
+    Chaining pull-through indices, having a pull-through point to another pull-through, does not
+    work.
+
 ## Use the newly created distribution
 
 The metadata and packages can now be retrieved from the distribution:

pulp_python/app/models.py

@@ -320,11 +320,25 @@ def get_remote_artifact_url(self, relative_path=None, request=None):
         """Get url for remote_artifact"""
         if request and (url := request.query.get("redirect")):
             # This is a special case for pull-through caching
+            # To handle PEP 658, it states that if the package has metadata available then it
+            # should be found at the download URL + ".metadata". Thus if the request url ends with
+            # ".metadata" then we need to add ".metadata" to the redirect url if not present.
+            if relative_path:
+                if relative_path.endswith(".metadata") and not url.endswith(".metadata"):
+                    url += ".metadata"
+                # Handle special case for bug in pip (TODO file issue in pip) where it appends
+                # ".metadata" to the redirect url instead of the request url
+                if url.endswith(".metadata") and not relative_path.endswith(".metadata"):
+                    setattr(self, "_real_relative_path", url.rsplit("/", 1)[1])
             return url
         return super().get_remote_artifact_url(relative_path, request=request)
 
     def get_remote_artifact_content_type(self, relative_path=None):
-        """Return PythonPackageContent."""
+        """Return PythonPackageContent, except for metadata artifacts."""
+        if hasattr(self, "_real_relative_path"):
+            relative_path = getattr(self, "_real_relative_path")
+        if relative_path and relative_path.endswith(".whl.metadata"):
+            return None
         return PythonPackageContent
 
     class Meta:

pulp_python/tests/functional/api/test_full_mirror.py

@@ -11,8 +11,9 @@
 
 from pypi_simple import ProjectPage
 from packaging.version import parse
-from urllib.parse import urljoin, urlsplit
+from urllib.parse import urljoin, urlsplit, urlunsplit
 from random import sample
+from hashlib import sha256
 
 
 def test_pull_through_install(
@@ -202,3 +203,38 @@ def test_pull_through_filtering_bad_names(python_remote_factory, python_distribu
     # Should have no packages with None version (they get filtered out)
     assert len(project_page.packages) > 0
     assert all(package.version is not None for package in project_page.packages)
+
+
+@pytest.mark.parallel
+def test_pull_through_metadata(python_remote_factory, python_distribution_factory):
+    """
+    Tests that metadata is correctly served when using pull-through.
+
+    So when requesting the metadata url according to PEP 658 you should just need to add .metadata
+    to the end of the url path. Since pull-through includes a redirect query parameter we need to
+    test adding .metadata to the end of the url path vs adding it to the end of redirect query.
+    """
+    remote = python_remote_factory(includes=["pytz"])
+    distro = python_distribution_factory(remote=remote.pulp_href)
+
+    url = f"{distro.base_url}simple/pytz/"
+    project_page = ProjectPage.from_response(requests.get(url), "pytz")
+    filename1 = "pytz-2023.2-py2.py3-none-any.whl"
+    filename2 = "pytz-2023.3-py2.py3-none-any.whl"
+    package1 = next(p for p in project_page.packages if p.filename == filename1)
+    package2 = next(p for p in project_page.packages if p.filename == filename2)
+    assert package1.has_metadata
+    assert package2.has_metadata
+
+    # The correct way to get the metadata url: add to path (uv does this)
+    parts1 = urlsplit(package1.url)
+    url1 = urlunsplit((parts1[0], parts1[1], parts1[2] + ".metadata", parts1[3], parts1[4]))
+    r = requests.get(url1)
+    assert r.status_code == 200
+    assert sha256(r.content).hexdigest() == package1.metadata_digests["sha256"]
+
+    # The incorrect way to get the metadata url: add to end of string (pip does this)
+    url2 = package2.url + ".metadata"
+    r = requests.get(url2)
+    assert r.status_code == 200
+    assert sha256(r.content).hexdigest() == package2.metadata_digests["sha256"]