Fix publish failure when handling bad dist name metadata

fixes: #907

Author: gerrod3

CHANGES/907.bugfix

@@ -0,0 +1 @@
+Fixed publication error when package's dists contain differing versions of its name.

pulp_python/app/tasks/publish.py

@@ -59,9 +59,9 @@ def write_simple_api(publication):
         python_models.PythonPackageContent.objects.filter(
             pk__in=publication.repository_version.content, _pulp_domain=domain
         )
-        .order_by("name")
+        .order_by("name__normalize")
         .values_list("name", flat=True)
-        .distinct()
+        .distinct("name__normalize")
     )
 
     # write the root index, which lists all of the projects for which there is a package available
@@ -80,22 +80,22 @@ def write_simple_api(publication):
     packages = python_models.PythonPackageContent.objects.filter(
         pk__in=publication.repository_version.content, _pulp_domain=domain
     )
-    releases = packages.order_by("name").values("name", "filename", "sha256")
+    releases = packages.order_by("name__normalize").values("name", "filename", "sha256")
 
     ind = 0
-    current_name = project_names[ind]
+    current_name = canonicalize_name(project_names[ind])
     package_releases = []
     for release in releases.iterator():
-        if release["name"] != current_name:
+        if canonicalize_name(release["name"]) != current_name:
             write_project_page(
-                name=canonicalize_name(current_name),
+                name=current_name,
                 simple_dir=simple_dir,
                 package_releases=package_releases,
                 publication=publication,
             )
             package_releases = []
             ind += 1
-            current_name = project_names[ind]
+            current_name = canonicalize_name(project_names[ind])
         relative_path = release["filename"]
         path = f"../../{relative_path}"
         checksum = release["sha256"]

pulp_python/pytest_plugin.py

@@ -178,7 +178,7 @@ def _gen_python_content(relative_path=PYTHON_EGG_FILENAME, url=None, **body):
 
         task = python_bindings.ContentPackagesApi.create(**body).task
         response = monitor_task(task)
-        return python_bindings.ContentPackagesApi.read(response.created_resources[0])
+        return python_bindings.ContentPackagesApi.read(response.created_resources[-1])
 
     yield _gen_python_content
 

pulp_python/tests/functional/api/test_crud_publications.py

@@ -1,5 +1,6 @@
 import pytest
 import random
+from pypi_simple import PyPISimple
 from urllib.parse import urljoin
 
 from pulp_python.tests.functional.constants import (
@@ -109,3 +110,29 @@ def test_new_content_is_published(python_publication_workflow, python_distributi
     url = urljoin(distro.base_url, "simple/")
     proper, msgs = ensure_simple(url, {"shelf-reader": releases})
     assert proper is True, msgs
+
+
+@pytest.mark.parallel
+def test_non_matching_canonicalized_name(
+    python_repo, python_content_factory, python_publication_factory, python_distribution_factory
+):
+    """Ensures a package with dists that have non-matching canonicalized names is published."""
+    packages = []
+    filenames = ["msg_parser-1.0.0-py2.py3-none-any.whl", "msg_parser-1.0.0.tar.gz"]
+    with PyPISimple() as client:
+        page = client.get_project_page("msg-parser")
+        for pkg in page.packages:
+            if pkg.filename in filenames:
+                c = python_content_factory(pkg.filename, url=pkg.url, repository=python_repo)
+                if c.filename.endswith(".tar.gz"):
+                    # The metadata name in the SDist is not the same as the Wheel's name
+                    assert c.name == "msg_parser"
+                else:
+                    assert c.name == "msg-parser"
+                packages.append(c)
+    pub = python_publication_factory(repository=python_repo)
+    distro = python_distribution_factory(publication=pub)
+
+    url = urljoin(distro.base_url, "simple/")
+    proper, msgs = ensure_simple(url, {"msg-parser": filenames})
+    assert proper is True, msgs