[PULP-715] WIP: Add JSON-based Simple API

pulp_python/app/migrations/0016_pythonpackagecontent_sha256_metadata_and_more.py

@@ -0,0 +1,30 @@
+# Generated by Django 4.2.24 on 2025-09-19 11:10
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+ dependencies = [
+ ("python", "0015_alter_pythonpackagecontent_options"),
+ ]
+
+ operations = [
+ migrations.AddField(
+ model_name="pythonpackagecontent",
+ name="sha256_metadata",
+ field=models.CharField(default="", max_length=64),
+ preserve_default=False,
+ ),
+ migrations.AddField(
+ model_name="pythonpackagecontent",
+ name="yanked",
+ field=models.BooleanField(default=False),
+ ),
+ migrations.AddField(
+ model_name="pythonpackagecontent",
+ name="yanked_reason",
+ field=models.TextField(default=""),
+ preserve_default=False,
+ ),
+ ]

pulp_python/app/models.py

@@ -192,6 +192,9 @@ class PythonPackageContent(Content):
  packagetype = models.TextField(choices=PACKAGE_TYPES)
  python_version = models.TextField()
  sha256 = models.CharField(db_index=True, max_length=64)
+ sha256_metadata = models.CharField(max_length=64)
+ yanked = models.BooleanField(default=False)
+ yanked_reason = models.TextField()
 
  # From pulpcore
  PROTECTED_FROM_RECLAIM = False

pulp_python/app/pypi/views.py

@@ -3,7 +3,9 @@
 
 from aiohttp.client_exceptions import ClientError
 from rest_framework.viewsets import ViewSet
+from rest_framework.renderers import BrowsableAPIRenderer, JSONRenderer, TemplateHTMLRenderer
 from rest_framework.response import Response
+from rest_framework.exceptions import NotAcceptable
 from django.core.exceptions import ObjectDoesNotExist
 from django.shortcuts import redirect
 from datetime import datetime, timezone, timedelta
@@ -43,7 +45,9 @@
 )
 from pulp_python.app.utils import (
  write_simple_index,
+ write_simple_index_json,
  write_simple_detail,
+ write_simple_detail_json,
  python_content_to_json,
  PYPI_LAST_SERIAL,
  PYPI_SERIAL_CONSTANT,
@@ -57,6 +61,17 @@
 ORIGIN_HOST = settings.CONTENT_ORIGIN if settings.CONTENT_ORIGIN else settings.PYPI_API_HOSTNAME
 BASE_CONTENT_URL = urljoin(ORIGIN_HOST, settings.CONTENT_PATH_PREFIX)
 
+PYPI_SIMPLE_V1_HTML = "application/vnd.pypi.simple.v1+html"
+PYPI_SIMPLE_V1_JSON = "application/vnd.pypi.simple.v1+json"
+
+
+class PyPISimpleHTMLRenderer(TemplateHTMLRenderer):
+ media_type = PYPI_SIMPLE_V1_HTML
+
+
+class PyPISimpleJSONRenderer(JSONRenderer):
+ media_type = PYPI_SIMPLE_V1_JSON
+
 
 class PyPIMixin:
  """Mixin to get index specific info."""
@@ -235,14 +250,42 @@ class SimpleView(PackageUploadMixin, ViewSet):
  ],
  }
 
+ def perform_content_negotiation(self, request, force=False):
+ """
+ Uses standard content negotiation, defaulting to HTML if no acceptable renderer is found.
+ """
+ try:
+ return super().perform_content_negotiation(request, force)
+ except NotAcceptable:
+ return TemplateHTMLRenderer(), TemplateHTMLRenderer.media_type # text/html
+
+ def get_renderers(self):
+ """
+ Uses custom renderers for PyPI Simple API endpoints, defaulting to standard ones.
+ """
+ if self.action in ["list", "retrieve"]:
+ # Ordered by priority if multiple content types are present
+ return [TemplateHTMLRenderer(), PyPISimpleHTMLRenderer(), PyPISimpleJSONRenderer()]
+ else:
+ return [JSONRenderer(), BrowsableAPIRenderer()]
+
  @extend_schema(summary="Get index simple page")
  def list(self, request, path):
  """Gets the simple api html page for the index."""
  repo_version, content = self.get_rvc()
  if self.should_redirect(repo_version=repo_version):
  return redirect(urljoin(self.base_content_url, f"{path}/simple/"))
  names = content.order_by("name").values_list("name", flat=True).distinct().iterator()
- return StreamingHttpResponse(write_simple_index(names, streamed=True))
+ media_type = request.accepted_renderer.media_type
+
+ if media_type == PYPI_SIMPLE_V1_JSON:
+ index_data = write_simple_index_json(names)
+ headers = {"X-PyPI-Last-Serial": str(PYPI_SERIAL_CONSTANT)}
+ return Response(index_data, headers=headers)
+ else:
+ index_data = write_simple_index(names, streamed=True)
+ kwargs = {"content_type": media_type}
+ return StreamingHttpResponse(index_data, **kwargs)
 
  def pull_through_package_simple(self, package, path, remote):
  """Gets the package's simple page from remote."""
@@ -252,7 +295,12 @@ def parse_package(release_package):
  stripped_url = urlunsplit(chain(parsed[:3], ("", "")))
  redirect_path = f"{path}/{release_package.filename}?redirect={stripped_url}"
  d_url = urljoin(self.base_content_url, redirect_path)
- return release_package.filename, d_url, release_package.digests.get("sha256", "")
+ return {
+ "filename": release_package.filename,
+ "url": d_url,
+ "sha256": release_package.digests.get("sha256", ""),
+ # todo: more fields?
+ }
 
  rfilter = get_remote_package_filter(remote)
  if not rfilter.filter_project(package):
@@ -269,7 +317,7 @@ def parse_package(release_package):
  except TimeoutException:
  return HttpResponse(f"{remote.url} timed out while fetching {package}.", status=504)
 
- if d.headers["content-type"] == "application/vnd.pypi.simple.v1+json":
+ if d.headers["content-type"] == PYPI_SIMPLE_V1_JSON:
  page = ProjectPage.from_json_data(json.load(open(d.path, "rb")), base_url=url)
  else:
  page = ProjectPage.from_html(package, open(d.path, "rb").read(), base_url=url)
@@ -290,7 +338,15 @@ def retrieve(self, request, path, package):
  return redirect(urljoin(self.base_content_url, f"{path}/simple/{normalized}/"))
  packages = (
  content.filter(name__normalize=normalized)
- .values_list("filename", "sha256", "name")
+ .values_list(
+ "filename",
+ "sha256",
+ "name",
+ "sha256_metadata",
+ "requires_python",
+ "yanked",
+ "yanked_reason",
+ )
  .iterator()
  )
  try:
@@ -300,8 +356,28 @@ def retrieve(self, request, path, package):
  else:
  packages = chain([present], packages)
  name = present[2]
- releases = ((f, urljoin(self.base_content_url, f"{path}/{f}"), d) for f, d, _ in packages)
- return StreamingHttpResponse(write_simple_detail(name, releases, streamed=True))
+ releases = (
+ {
+ "filename": f,
+ "url": urljoin(self.base_content_url, f"{path}/{f}"),
+ "sha256": s,
+ "sha256_metadata": sm,
+ "requires_python": rp,
+ "yanked": y,
+ "yanked_reason": yr,
+ }
+ for f, s, _, sm, rp, y, yr in packages
+ )
+ media_type = request.accepted_renderer.media_type
+
+ if media_type == PYPI_SIMPLE_V1_JSON:
+ detail_data = write_simple_detail_json(name, releases)
+ headers = {"X-PyPI-Last-Serial": str(PYPI_SERIAL_CONSTANT)}
+ return Response(detail_data, headers=headers)
+ else:
+ detail_data = write_simple_detail(name, releases, streamed=True)
+ kwargs = {"content_type": media_type}
+ return StreamingHttpResponse(detail_data, kwargs)
 
  @extend_schema(
  request=PackageUploadSerializer,

pulp_python/app/tasks/publish.py

@@ -118,6 +118,7 @@ def write_project_page(name, simple_dir, package_releases, publication):
  metadata_relative_path = f"{project_dir}index.html"
 
  with open(metadata_relative_path, "w") as simple_metadata:
+ # todo?
  simple_metadata.write(write_simple_detail(name, package_releases))
 
  project_metadata = models.PublishedMetadata.create_from_file(

pulp_python/app/utils.py

@@ -12,10 +12,13 @@
 from pulpcore.plugin.models import Remote
 
 
+# todo: why upper case?
 PYPI_LAST_SERIAL = "X-PYPI-LAST-SERIAL"
 """TODO This serial constant is temporary until Python repositories implements serials"""
 PYPI_SERIAL_CONSTANT = 1000000000
 
+PYPI_API_VERSION = "1.0"
+
 simple_index_template = """<!DOCTYPE html>
 <html>
  <head>
@@ -38,8 +41,8 @@
 </head>
 <body>
  <h1>Links for {{ project_name }}</h1>
- {% for name, path, sha256 in project_packages %}
- <a href="{{ path }}#sha256={{ sha256 }}" rel="internal">{{ name }}</a><br/>
+ {% for pkg in project_packages %}
+ <a href="{{ pkg.url }}#sha256={{ pkg.sha256 }}" rel="internal">{{ pkg.filename }}</a><br/>
  {% endfor %}
 </body>
 </html>
@@ -158,6 +161,9 @@ def parse_metadata(project, version, distribution):
  package["requires_python"] = distribution.get("requires_python") or package.get(
  "requires_python"
  ) # noqa: E501
+ package["yanked"] = distribution.get("yanked") or False
+ package["yanked_reason"] = distribution.get("yanked_reason") or ""
+ package["sha256_metadata"] = distribution.get("data-dist-info-metadata", {}).get("sha256") or ""
 
  return package
 
@@ -203,6 +209,10 @@ def artifact_to_python_content_data(filename, artifact, domain=None):
  data["filename"] = filename
  data["pulp_domain"] = domain or artifact.pulp_domain
  data["_pulp_domain"] = data["pulp_domain"]
+ # todo: how to get these / should they be here?
+ # data["yanked"] = False
+ # data["yanked_reason"] = ""
+ # data["sha256_metadata"] = ""
  return data
 
 
@@ -325,6 +335,7 @@ def python_content_to_info(content):
  "platform": content.platform or "",
  "requires_dist": json_to_dict(content.requires_dist) or None,
  "classifiers": json_to_dict(content.classifiers) or None,
+ # todo yanked
  "yanked": False, # These are no longer used on PyPI, but are still present
  "yanked_reason": None,
  # New core metadata (Version 2.1, 2.2, 2.4)
@@ -395,6 +406,7 @@ def find_artifact():
  "upload_time": str(content.pulp_created),
  "upload_time_iso_8601": str(content.pulp_created.isoformat()),
  "url": url,
+ # todo yanked
  "yanked": False,
  "yanked_reason": None,
  }
@@ -414,6 +426,53 @@ def write_simple_detail(project_name, project_packages, streamed=False):
  return detail.stream(**context) if streamed else detail.render(**context)
 
 
+def write_simple_index_json(project_names):
+ """Writes the simple index in JSON format."""
+ return {
+ "meta": {"api-version": PYPI_API_VERSION, "_last-serial": PYPI_SERIAL_CONSTANT},
+ "projects": [
+ {"name": name, "_last-serial": PYPI_SERIAL_CONSTANT} for name in project_names
+ ],
+ }
+
+
+def write_simple_detail_json(project_name, project_packages):
+ """Writes the simple detail page in JSON format."""
+ return {
+ "meta": {"api-version": PYPI_API_VERSION, "_last-serial": PYPI_SERIAL_CONSTANT},
+ "name": canonicalize_name(project_name),
+ "files": [
+ {
+ # v1.0, PEP 691
+ "filename": package["filename"],
+ "url": package["url"],
+ "hashes": {"sha256": package["sha256"]},
+ "requires_python": package["requires_python"] or None,
+ # data-dist-info-metadata is deprecated alias for core-metadata
+ "data-dist-info-metadata": (
+ {"sha256": package["sha256_metadata"]} if package["sha256_metadata"] else False
+ ),
+ "yanked": (
+ package["yanked_reason"]
+ if package["yanked"] and package["yanked_reason"]
+ else package["yanked"]
+ ),
+ # gpg-sig (not in warehouse)
+ # todo (from new PEPs):
+ # size (v1.1, PEP 700)
+ # upload-time (v1.1, PEP 700)
+ # core-metadata (PEP 7.14)
+ # provenance (v1.3, PEP 740)
+ }
+ for package in project_packages
+ ],
+ # todo (from new PEPs):
+ # versions (v1.1, PEP 700)
+ # alternate-locations (v1.2, PEP 708)
+ # project-status (v1.4, PEP 792 - pypi and docs differ)
+ }
+
+
 class PackageIncludeFilter:
  """A special class to help filter Package's based on a remote's include/exclude"""
 

pulp_python/tests/functional/api/test_full_mirror.py

@@ -66,7 +66,7 @@ def test_pull_through_filter(python_remote_factory, python_distribution_factory)
 
  r = requests.get(f"{distro.base_url}simple/pulpcore/")
  assert r.status_code == 404
- assert r.json() == {"detail": "pulpcore does not exist."}
+ assert r.text == "404 Not Found"
 
  r = requests.get(f"{distro.base_url}simple/shelf-reader/")
  assert r.status_code == 200
@@ -86,7 +86,7 @@ def test_pull_through_filter(python_remote_factory, python_distribution_factory)
 
  r = requests.get(f"{distro.base_url}simple/django/")
  assert r.status_code == 404
- assert r.json() == {"detail": "django does not exist."}
+ assert r.text == "404 Not Found"
 
  r = requests.get(f"{distro.base_url}simple/pulpcore/")
  assert r.status_code == 502