From 8fcd1530e2dfca733f98c1c555bfd1ed77e19d7d Mon Sep 17 00:00:00 2001
From: Gerrod Ubben <gerrodubben@gmail.com>
Date: Mon, 22 Nov 2021 17:43:18 -0500
Subject: [PATCH] Fix name normalization for package simple detail view

fixes: #467
---
 CHANGES/467.bugfix            |  1 +
 pulp_python/app/models.py     |  9 +++++++++
 pulp_python/app/pypi/views.py | 24 ++++++++++++++++++------
 3 files changed, 28 insertions(+), 6 deletions(-)
 create mode 100644 CHANGES/467.bugfix

diff --git a/CHANGES/467.bugfix b/CHANGES/467.bugfix
new file mode 100644
index 000000000..3a5209199
--- /dev/null
+++ b/CHANGES/467.bugfix
@@ -0,0 +1 @@
+Fixed package name normalization issue preventing installing packages with "." or "_" in their names.
diff --git a/pulp_python/app/models.py b/pulp_python/app/models.py
index 0f10b2e75..fbefb4c12 100644
--- a/pulp_python/app/models.py
+++ b/pulp_python/app/models.py
@@ -124,6 +124,14 @@ class Meta:
         default_related_name = "%(app_label)s_%(model_name)s"
 
 
+class NormalizeName(models.Transform):
+    """A transform field to normalize package names according to PEP426."""
+
+    function = "REGEXP_REPLACE"
+    template = "LOWER(%(function)s(%(expressions)s, '(\.|_|-)', '-', 'ig'))"  # noqa:W605
+    lookup_name = "normalize"
+
+
 class PythonPackageContent(Content):
     """
     A Content Type representing Python's Distribution Package.
@@ -142,6 +150,7 @@ class PythonPackageContent(Content):
     filename = models.TextField(db_index=True)
     packagetype = models.TextField(choices=PACKAGE_TYPES)
     name = models.TextField()
+    name.register_lookup(NormalizeName)
     version = models.TextField()
     sha256 = models.CharField(unique=True, db_index=True, max_length=64)
     # Optional metadata
diff --git a/pulp_python/app/pypi/views.py b/pulp_python/app/pypi/views.py
index c3020b7cc..70a3f2439 100644
--- a/pulp_python/app/pypi/views.py
+++ b/pulp_python/app/pypi/views.py
@@ -19,8 +19,10 @@
 )
 from drf_spectacular.utils import extend_schema
 from dynaconf import settings
+from itertools import chain
 from urllib.parse import urljoin
 from pathlib import PurePath
+from packaging.utils import canonicalize_name
 
 from pulpcore.plugin.tasking import dispatch
 from pulp_python.app.models import (
@@ -188,14 +190,24 @@ def list(self, request, path):
     def retrieve(self, request, path, package):
         """Retrieves the simple api html page for a package."""
         distro, repo_ver, content = self.get_drvc(path)
+        # Should I redirect if the normalized name is different?
+        normalized = canonicalize_name(package)
         if self.should_redirect(distro, repo_version=repo_ver):
-            # Maybe this name needs to be normalized?
-            return redirect(urljoin(BASE_CONTENT_URL, f'{path}/simple/{package}/'))
-        packages = content.filter(name__iexact=package).values_list('filename', 'sha256').iterator()
-        detail_packages = ((f, urljoin(BASE_CONTENT_URL, f'{path}/{f}'), d) for f, d in packages)
-        return StreamingHttpResponse(
-            write_simple_detail(package, detail_packages, streamed=True)
+            return redirect(urljoin(BASE_CONTENT_URL, f'{path}/simple/{normalized}/'))
+        packages = (
+            content.filter(name__normalize=normalized)
+            .values_list('filename', 'sha256', 'name')
+            .iterator()
         )
+        try:
+            present = next(packages)
+        except StopIteration:
+            raise Http404(f"{normalized} does not exist.")
+        else:
+            packages = chain([present], packages)
+            name = present[2]
+        releases = ((f, urljoin(BASE_CONTENT_URL, f'{path}/{f}'), d) for f, d, _ in packages)
+        return StreamingHttpResponse(write_simple_detail(name, releases, streamed=True))
 
     @extend_schema(request=PackageUploadSerializer,
                    responses={200: PackageUploadTaskSerializer},