Fix incorrect Content-Type header in Registry API

fixes: #1997
(cherry picked from commit e416df5)

Author: gerrod3

CHANGES/1997.bugfix

@@ -0,0 +1 @@
+Fixed the Registry API returning 'charset utf-8' in the Content-Type header.

pulp_container/app/cache.py

@@ -37,6 +37,13 @@ def make_key(self, request):
         key = ":".join(all_keys[k] for k in self.keys)
         return key
 
+    async def make_response(self, key, base_key):
+        """Fix the Content-Type header to remove the charset."""
+        response = await super().make_response(key, base_key)
+        if response is not None:
+            response.headers["Content-Type"] = response.headers["Content-Type"].split(";")[0]
+        return response
+
 
 class RegistryApiCache(RegistryCache, SyncContentCache):
     """A wrapper around the Redis content cache handler tailored for the registry API."""

pulp_container/app/registry.py

@@ -140,7 +140,7 @@ async def get_tag(self, request):
                     "Docker-Content-Digest": digest,
                     "Docker-Distribution-API-Version": "registry/2.0",
                 }
-                return web.Response(text=raw_text_manifest, headers=headers)
+                return web.Response(body=raw_text_manifest, headers=headers)
             else:
                 raise PathNotResolved(tag_name)
 
@@ -176,7 +176,7 @@ async def get_tag(self, request):
                         "Docker-Content-Digest": digest,
                         "Docker-Distribution-API-Version": "registry/2.0",
                     }
-                    return web.Response(text=raw_text_manifest, headers=headers)
+                    return web.Response(body=raw_text_manifest, headers=headers)
 
         accepted_media_types = get_accepted_media_types(request.headers)
 
@@ -200,7 +200,7 @@ async def get_tag(self, request):
                 "Content-Type": return_media_type,
                 "Docker-Content-Digest": tag.tagged_manifest.digest,
             }
-            return web.Response(text=tag.tagged_manifest.data, headers=response_headers)
+            return web.Response(body=tag.tagged_manifest.data, headers=response_headers)
 
         # return what was found in case media_type is accepted header (docker, oci)
         if tag.tagged_manifest.media_type in accepted_media_types:
@@ -209,7 +209,7 @@ async def get_tag(self, request):
                 "Content-Type": return_media_type,
                 "Docker-Content-Digest": tag.tagged_manifest.digest,
             }
-            return web.Response(text=tag.tagged_manifest.data, headers=response_headers)
+            return web.Response(body=tag.tagged_manifest.data, headers=response_headers)
 
         # return 404 in case the client is requesting docker manifest v2 schema 1
         raise PathNotResolved(tag_name)
@@ -249,7 +249,7 @@ async def get_by_digest(self, request):
                     "Content-Type": manifest.media_type,
                     "Docker-Content-Digest": manifest.digest,
                 }
-                return web.Response(text=manifest.data, headers=headers)
+                return web.Response(body=manifest.data, headers=headers)
             elif content_type == "blobs":
                 ca = await ContentArtifact.objects.select_related("artifact", "content").aget(
                     content__in=content, relative_path=digest
@@ -278,7 +278,7 @@ async def get_by_digest(self, request):
                         "Docker-Content-Digest": digest,
                         "Docker-Distribution-API-Version": "registry/2.0",
                     }
-                    return web.Response(text=raw_text_manifest, headers=headers)
+                    return web.Response(body=raw_text_manifest, headers=headers)
                 elif content_type == "blobs":
                     # there might be a case where the client has all the manifest data in place
                     # and tries to download only missing blobs; because of that, only the reference

pulp_container/tests/functional/api/test_content_cache.py

@@ -31,23 +31,26 @@ def _check_content(expect_metadata, dist_url, tag_name="latest", headers=None):
         duplicated_content_units = sorted([manifest_by_tag, manifest_by_digest] * 2)
         for i, c in enumerate(duplicated_content_units):
             url = urljoin(dist_url, c)
-            check_cache(url, expect_metadata(i), headers, auth)
+            check_cache(url, expect_metadata(i), headers, auth, MEDIA_TYPE.MANIFEST_LIST)
 
         duplicated_content_units = [blob_by_digest] * 2
         for i, c in enumerate(duplicated_content_units):
             url = urljoin(dist_url, c)
-            check_cache(url, expect_metadata(i), headers, auth)
+            check_cache(url, expect_metadata(i), headers, auth, "application/octet-stream")
 
     return _check_content
 
 
-def check_cache(url, expected_metadata, headers, auth):
+def check_cache(url, expected_metadata, headers, auth, content_type):
     """A helper function to check if cache miss or hit occurred."""
     auth = get_auth_for_url(url, auth=auth)
 
     response = requests.get(url, auth=auth, headers=headers)
     response_metadata = fetch_response_metadata(response)
     assert expected_metadata == response_metadata, url
+    # Check that we return the correct content type for both cache hits and misses
+    if response.status_code == 200:
+        assert response.headers.get("Content-Type") == content_type, url
 
 
 def fetch_response_metadata(response):