PEP 740: add provenance to simple API (#16801)

Co-authored-by: Mike Fiedler <[email protected]>

tests/functional/api/test_simple.py

@@ -12,7 +12,12 @@
 
 from http import HTTPStatus
 
-from ...common.db.packaging import FileFactory, ProjectFactory, ReleaseFactory
+from ...common.db.packaging import (
+    FileFactory,
+    ProjectFactory,
+    ProvenanceFactory,
+    ReleaseFactory,
+)
 
 
 def test_simple_api_html(webtest):
@@ -34,3 +39,24 @@ def test_simple_api_detail(webtest):
     assert resp.html.h1.string == f"Links for {project.normalized_name}"
     # There should be a link for every file
     assert len(resp.html.find_all("a")) == 2
+
+
+def test_simple_api_has_provenance(webtest):
+    project = ProjectFactory.create()
+    release = ReleaseFactory.create(project=project)
+    files = FileFactory.create_batch(2, release=release, packagetype="bdist_wheel")
+
+    for file in files:
+        ProvenanceFactory.create(file=file)
+
+    resp = webtest.get(f"/simple/{project.normalized_name}/", status=HTTPStatus.OK)
+    links = resp.html.find_all("a")
+
+    for file in files:
+        link = next(link for link in links if link.text == file.filename)
+        provenance_url = link.get("data-provenance")
+
+        assert provenance_url == (
+            f"http://localhost/integrity/{file.release.project.normalized_name}/"
+            f"{file.release.version}/{file.filename}/provenance"
+        )

tests/functional/forklift/test_legacy.py

@@ -329,3 +329,14 @@ def test_provenance_upload(webtest):
     assert len(attestations) == 1
     attestation = attestations[0]
     assert attestation == json.loads(attestation_contents)
+
+    # While we needed to be authenticated to upload a project, this is no longer
+    # required to view it.
+    webtest.authorization = None
+    expected_filename = "sampleproject-3.0.0.tar.gz"
+
+    response = webtest.get(
+        f"/integrity/{project.name}/3.0.0/{expected_filename}/provenance",
+        status=HTTPStatus.OK,
+    )
+    assert response.json == project.releases[0].files[0].provenance.provenance

tests/unit/api/test_simple.py

@@ -17,6 +17,7 @@
 from pyramid.httpexceptions import HTTPMovedPermanently
 from pyramid.testing import DummyRequest
 
+from tests.common.db.oidc import GitHubPublisherFactory
 from warehouse.api import simple
 from warehouse.packaging.utils import API_VERSION, _valid_simple_detail_context
 
@@ -26,6 +27,7 @@
     FileFactory,
     JournalEntryFactory,
     ProjectFactory,
+    ProvenanceFactory,
     ReleaseFactory,
 )
 
@@ -298,6 +300,7 @@ def test_with_files_no_serial(self, db_request, content_type, renderer_override)
                     "upload-time": f.upload_time.isoformat() + "Z",
                     "data-dist-info-metadata": False,
                     "core-metadata": False,
+                    "provenance": None,
                 }
                 for f in files
             ],
@@ -349,6 +352,7 @@ def test_with_files_with_serial(self, db_request, content_type, renderer_overrid
                     "upload-time": f.upload_time.isoformat() + "Z",
                     "data-dist-info-metadata": False,
                     "core-metadata": False,
+                    "provenance": None,
                 }
                 for f in files
             ],
@@ -445,6 +449,7 @@ def test_with_files_with_version_multi_digit(
                         if f.metadata_file_sha256_digest is not None
                         else False
                     ),
+                    "provenance": None,
                 }
                 for f in files
             ],
@@ -489,6 +494,126 @@ def test_with_files_quarantined_omitted_from_index(
         if renderer_override is not None:
             assert db_request.override_renderer == renderer_override
 
+    @pytest.mark.parametrize(
+        ("content_type", "renderer_override"),
+        CONTENT_TYPE_PARAMS,
+    )
+    def test_with_files_varying_provenance(
+        self,
+        db_request,
+        integrity_service,
+        dummy_attestation,
+        content_type,
+        renderer_override,
+    ):
+        db_request.accept = content_type
+        db_request.oidc_publisher = GitHubPublisherFactory.create()
+
+        project = ProjectFactory.create()
+        release = ReleaseFactory.create(project=project, version="1.0.0")
+
+        # wheel with provenance, sdist with no provenance
+        wheel = FileFactory.create(
+            release=release,
+            filename=f"{project.name}-1.0.0.whl",
+            packagetype="bdist_wheel",
+            metadata_file_sha256_digest="deadbeefdeadbeefdeadbeefdeadbeef",
+        )
+
+        provenance = ProvenanceFactory.create(file=wheel)
+        assert wheel.provenance == provenance
+
+        sdist = FileFactory.create(
+            release=release,
+            filename=f"{project.name}-1.0.0.tar.gz",
+            packagetype="sdist",
+        )
+
+        files = [sdist, wheel]
+
+        db_request.matchdict["name"] = project.normalized_name
+
+        def route_url(route, **kw):
+            # Rendering the simple index calls route_url once for each file,
+            # and zero or one times per file depending on whether the file
+            # has provenance. We emulate this while testing by maintaining
+            # a dictionary of expected URLs for each route, which are
+            # pulled from as appropriate when a route_url call is made.
+            route_urls = {
+                "packaging.file": {f.path: f"/file/{f.filename}" for f in files},
+                "integrity.provenance": {
+                    (
+                        wheel.release.project.normalized_name,
+                        wheel.release.version,
+                        wheel.filename,
+                    ): (
+                        f"/integrity/{wheel.release.project.normalized_name}/"
+                        f"{wheel.release.version}/{wheel.filename}/provenance"
+                    )
+                },
+            }
+
+            match route:
+                case "packaging.file":
+                    return route_urls[route].get(kw.get("path"), "")
+                case "integrity.provenance":
+                    key = (
+                        kw.get("project_name"),
+                        kw.get("release"),
+                        kw.get("filename"),
+                    )
+                    return route_urls[route].get(key, "")
+                case _:
+                    pytest.fail(f"unexpected route: {route}")
+
+        db_request.route_url = route_url
+
+        user = UserFactory.create()
+        je = JournalEntryFactory.create(name=project.name, submitted_by=user)
+
+        context = {
+            "meta": {"_last-serial": je.id, "api-version": API_VERSION},
+            "name": project.normalized_name,
+            "versions": ["1.0.0"],
+            "files": [
+                {
+                    "filename": f.filename,
+                    "url": f"/file/{f.filename}",
+                    "hashes": {"sha256": f.sha256_digest},
+                    "requires-python": f.requires_python,
+                    "yanked": False,
+                    "size": f.size,
+                    "upload-time": f.upload_time.isoformat() + "Z",
+                    "data-dist-info-metadata": (
+                        {"sha256": "deadbeefdeadbeefdeadbeefdeadbeef"}
+                        if f.metadata_file_sha256_digest is not None
+                        else False
+                    ),
+                    "core-metadata": (
+                        {"sha256": "deadbeefdeadbeefdeadbeefdeadbeef"}
+                        if f.metadata_file_sha256_digest is not None
+                        else False
+                    ),
+                    "provenance": (
+                        (
+                            f"/integrity/{f.release.project.normalized_name}/"
+                            f"{f.release.version}/{f.filename}/provenance"
+                        )
+                        if f.provenance is not None
+                        else None
+                    ),
+                }
+                for f in files
+            ],
+            "alternate-locations": [],
+        }
+        context = _update_context(context, content_type, renderer_override)
+
+        assert simple.simple_detail(project, db_request) == context
+
+        if renderer_override is not None:
+            assert db_request.override_renderer == renderer_override
+
 
 def _update_context(context, content_type, renderer_override):
     if renderer_override != "json" or content_type in [

warehouse/packaging/utils.py

@@ -22,7 +22,7 @@
 from warehouse.packaging.interfaces import ISimpleStorage
 from warehouse.packaging.models import File, LifecycleStatus, Project, Release
 
-API_VERSION = "1.2"
+API_VERSION = "1.3"
 
 
 def _simple_index(request, serial):
@@ -101,6 +101,16 @@ def _simple_detail(project, request):
                     if file.metadata_file_sha256_digest
                     else False
                 ),
+                "provenance": (
+                    request.route_url(
+                        "integrity.provenance",
+                        project_name=project.normalized_name,
+                        release=file.release.version,
+                        filename=file.filename,
+                    )
+                    if file.provenance
+                    else None
+                ),
             }
             for file in files
         ],

warehouse/templates/api/simple/detail.html

@@ -23,7 +23,7 @@
   <body>
     <h1>Links for {{ name }}</h1>
     {% for file in files -%}
-    <a href="{{ file.url }}#sha256={{ file.hashes.sha256 }}" {% if file.get('requires-python') %}data-requires-python="{{ file['requires-python'] }}" {% endif %}{% if file.yanked %}data-yanked="{% if file.yanked is string %}{{ file.yanked }}{% endif %}" {% endif %}{% if file['core-metadata'] %}data-dist-info-metadata="sha256={{ file['core-metadata']['sha256'] }}" data-core-metadata="sha256={{ file['core-metadata']['sha256'] }}"{% endif %}>{{ file.filename }}</a><br />
+    <a href="{{ file.url }}#sha256={{ file.hashes.sha256 }}" {% if file.get('requires-python') %}data-requires-python="{{ file['requires-python'] }}" {% endif %}{% if file.yanked %}data-yanked="{% if file.yanked is string %}{{ file.yanked }}{% endif %}" {% endif %}{% if file['core-metadata'] %}data-dist-info-metadata="sha256={{ file['core-metadata']['sha256'] }}" data-core-metadata="sha256={{ file['core-metadata']['sha256'] }}"{% endif %}{% if file.get('provenance') %}data-provenance="{{ file['provenance'] }}" {% endif %}>{{ file.filename }}</a><br />
     {% endfor -%}
   </body>
 </html>