Fully support the latest core metadata (up to 2.4)

closes #689

Author: jobselko

CHANGES/689.feature

@@ -0,0 +1 @@
+Added full support for the latest core metadata (up to 2.4).

pulp_python/app/migrations/0014_pythonpackagecontent_dynamic_and_more.py

@@ -0,0 +1,33 @@
+# Generated by Django 4.2.19 on 2025-07-02 13:44
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('python', '0013_add_rbac_permissions'),
+    ]
+
+    operations = [
+        migrations.AddField(
+            model_name='pythonpackagecontent',
+            name='dynamic',
+            field=models.JSONField(default=list),
+        ),
+        migrations.AddField(
+            model_name='pythonpackagecontent',
+            name='license_expression',
+            field=models.TextField(default=''),
+        ),
+        migrations.AddField(
+            model_name='pythonpackagecontent',
+            name='license_file',
+            field=models.JSONField(default=list),
+        ),
+        migrations.AddField(
+            model_name='pythonpackagecontent',
+            name='provides_extra',
+            field=models.JSONField(default=list),
+        ),
+    ]

pulp_python/app/models.py

@@ -140,10 +140,15 @@ class PythonPackageContent(Content):
     """
     A Content Type representing Python's Distribution Package.
 
-    As defined in pep-0426 and pep-0345.
+    Core Metadata:
+        https://packaging.python.org/en/latest/specifications/core-metadata/
 
-    https://www.python.org/dev/peps/pep-0491/
-    https://www.python.org/dev/peps/pep-0345/
+    Release metadata (JSON API):
+        https://docs.pypi.org/api/json/
+
+    File Formats:
+        https://packaging.python.org/en/latest/specifications/source-distribution-format/
+        https://packaging.python.org/en/latest/specifications/binary-distribution-format/
     """
     # Core metadata
     # Version 1.0
@@ -174,6 +179,12 @@ class PythonPackageContent(Content):
     requires_python = models.TextField()
     # Version 2.1
     description_content_type = models.TextField()
+    provides_extra = models.JSONField(default=list)
+    # Version 2.2
+    dynamic = models.JSONField(default=list)
+    # Version 2.4
+    license_expression = models.TextField(default="")
+    license_file = models.JSONField(default=list)
 
     # Release metadata
     filename = models.TextField(db_index=True)

pulp_python/app/serializers.py

@@ -184,6 +184,26 @@ class PythonPackageContentSerializer(core_serializers.SingleArtifactContentUploa
         help_text=_('A string stating the markup syntax (if any) used in the distribution’s'
                     ' description, so that tools can intelligently render the description.')
     )
+    provides_extra = serializers.JSONField(
+        required=False, default=list,
+        help_text=_('A JSON list containing names of optional features provided by the package.')
+    )
+    # Version 2.2
+    dynamic = serializers.JSONField(
+        required=False, default=list,
+        help_text=_('A JSON list containing names of other core metadata fields which are '
+                    'permitted to vary between sdist and bdist packages. Fields NOT marked '
+                    'dynamic MUST be the same between bdist and sdist.')
+    )
+    # Version 2.4
+    license_expression = serializers.CharField(
+        required=False, allow_blank=True,
+        help_text=_('Text string that is a valid SPDX license expression.')
+    )
+    license_file = serializers.JSONField(
+        required=False, default=list,
+        help_text=_('A JSON list containing names of the paths to license-related files.')
+    )
     # Release metadata
     filename = serializers.CharField(
         help_text=_('The name of the distribution package, usually of the format:'
@@ -258,6 +278,7 @@ class Meta:
             'download_url', 'supported_platform', 'maintainer', 'maintainer_email',
             'obsoletes_dist', 'project_url', 'project_urls', 'provides_dist', 'requires_external',
             'requires_dist', 'requires_python', 'description_content_type',
+            'provides_extra', 'dynamic', 'license_expression', 'license_file',
             'filename', 'packagetype', 'python_version', 'sha256'
         )
         model = python_models.PythonPackageContent

pulp_python/app/utils.py

@@ -117,6 +117,12 @@ def parse_project_metadata(project):
         'requires_python': project.get('requires_python') or "",
         # Version 2.1
         'description_content_type': project.get('description_content_type') or "",
+        'provides_extra': json.dumps(project.get('provides_extra', [])),
+        # Version 2.2
+        'dynamic': json.dumps(project.get('dynamic', [])),
+        # Version 2.4
+        'license_expression': project.get('license_expression') or "",
+        'license_file': json.dumps(project.get('license_file', [])),
         # Release metadata
         'packagetype': project.get('packagetype') or "",
         'python_version': project.get('python_version') or "",
@@ -316,6 +322,11 @@ def python_content_to_info(content):
         "classifiers": json_to_dict(content.classifiers) or None,
         "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)
+        "provides_extra": json_to_dict(content.provides_extra) or None,
+        "dynamic": json_to_dict(content.dynamic) or None,
+        "license_expression": content.license_expression or "",
+        "license_file": json_to_dict(content.license_file) or None,
     }
 
 

pulp_python/tests/functional/api/test_crud_content_unit.py

@@ -110,6 +110,33 @@ def test_content_crud(
     msg = "The uploaded artifact's sha256 checksum does not match the one provided"
     assert msg in e.value.task.error["description"]
 
+def test_content_create_new_metadata(
+    delete_orphans_pre, download_python_file, monitor_task, python_bindings
+):
+    """
+    Test the creation of python content unit with newly added core metadata (provides_extra,
+    dynamic, license_expression, license_file).
+    """
+    python_egg_filename = "setuptools-80.9.0.tar.gz"
+    python_egg_url = urljoin(urljoin(PYTHON_FIXTURES_URL, "packages/"), python_egg_filename)
+    python_file = download_python_file(python_egg_filename, python_egg_url)
+
+    body = {"relative_path": python_egg_filename, "file": python_file}
+    response = python_bindings.ContentPackagesApi.create(**body)
+    task = monitor_task(response.task)
+    content = python_bindings.ContentPackagesApi.read(task.created_resources[0])
+
+    python_package_data = {
+        "filename": "setuptools-80.9.0.tar.gz",
+        # todo: "provides_extra" should be:
+        #  '["test", "doc", "ssl", "certs", "core", "check", "cover", "enabler", "type"]'
+        "provides_extra": "[]",
+        "dynamic": '["license-file"]',
+        "license_expression": "MIT",
+        "license_file": '["LICENSE"]',
+    }
+    for k, v in python_package_data.items():
+        assert getattr(content, k) == v
 
 @pytest.mark.parallel
 def test_upload_metadata_23_spec(python_content_factory):
@@ -139,11 +166,13 @@ def test_upload_requires_python(python_content_factory):
 @pytest.mark.parallel
 def test_upload_metadata_24_spec(python_content_factory):
     """Test that packages using metadata spec 2.4 can be uploaded to pulp."""
-    filename = "urllib3-2.3.0-py3-none-any.whl"
+    filename = "setuptools-80.9.0.tar.gz"
     with PyPISimple() as client:
-        page = client.get_project_page("urllib3")
+        page = client.get_project_page("setuptools")
         for package in page.packages:
             if package.filename == filename:
                 content = python_content_factory(filename, url=package.url)
                 assert content.metadata_version == "2.4"
+                assert content.license_expression == "MIT"
+                assert content.license_file == '["LICENSE"]'
                 break