From 7f068c3fc37d618940a50d5c8b80535d041edcb6 Mon Sep 17 00:00:00 2001
From: Joseph Schorr <joseph.schorr@coreos.com>
Date: Fri, 11 Jan 2019 15:24:21 -0500
Subject: [PATCH] Skip parsing the manifest where applicable

Instead of always parsing (like we did previously), we now only parse the manifest if conversion is necessary. This should save significant CPU.
---
 endpoints/v2/manifest.py | 54 ++++++++++++++++++----------------------
 1 file changed, 24 insertions(+), 30 deletions(-)

diff --git a/endpoints/v2/manifest.py b/endpoints/v2/manifest.py
index 75f818736..4ed8331fd 100644
--- a/endpoints/v2/manifest.py
+++ b/endpoints/v2/manifest.py
@@ -57,15 +57,9 @@ def fetch_manifest_by_tagname(namespace_name, repo_name, manifest_ref):
     # Something went wrong.
     raise ManifestInvalid()
 
-  try:
-    parsed = manifest.get_parsed_manifest()
-  except ManifestException:
-    logger.exception('Got exception when trying to parse manifest `%s`', manifest_ref)
-    raise ManifestInvalid()
-
-  supported = _rewrite_schema_if_necessary(namespace_name, repo_name, manifest_ref, manifest,
-                                           parsed)
-  if supported is None:
+  manifest_bytes, manifest_digest, manifest_media_type = _rewrite_schema_if_necessary(
+    namespace_name, repo_name, manifest_ref, manifest)
+  if manifest_bytes is None:
     raise ManifestUnknown()
 
   track_and_log('pull_repo', repository_ref, analytics_name='pull_repo_100x', analytics_sample=0.01,
@@ -73,11 +67,11 @@ def fetch_manifest_by_tagname(namespace_name, repo_name, manifest_ref):
   metric_queue.repository_pull.Inc(labelvalues=[namespace_name, repo_name, 'v2', True])
 
   return Response(
-    supported.bytes.as_unicode(),
+    manifest_bytes.as_unicode(),
     status=200,
     headers={
-      'Content-Type': '%s; charset=utf-8' % supported.media_type,
-      'Docker-Content-Digest': supported.digest,
+      'Content-Type': '%s; charset=utf-8' % manifest_media_type,
+      'Docker-Content-Digest': manifest_digest,
     },
   )
 
@@ -96,43 +90,43 @@ def fetch_manifest_by_digest(namespace_name, repo_name, manifest_ref):
   if manifest is None:
     raise ManifestUnknown()
 
-  try:
-    parsed = manifest.get_parsed_manifest()
-  except ManifestException:
-    logger.exception('Got exception when trying to parse manifest `%s`', manifest_ref)
-    raise ManifestInvalid()
-
-  supported = _rewrite_schema_if_necessary(namespace_name, repo_name, '$digest', manifest,
-                                           parsed)
-  if supported is None:
+  manifest_bytes, manifest_digest, manifest_media_type = _rewrite_schema_if_necessary(
+    namespace_name, repo_name, '$digest', manifest)
+  if manifest_digest is None:
     raise ManifestUnknown()
 
   track_and_log('pull_repo', repository_ref, manifest_digest=manifest_ref)
   metric_queue.repository_pull.Inc(labelvalues=[namespace_name, repo_name, 'v2', True])
 
-  return Response(supported.bytes.as_unicode(), status=200, headers={
-    'Content-Type': '%s; charset=utf-8' % supported.media_type,
-    'Docker-Content-Digest': supported.digest,
+  return Response(manifest_bytes.as_unicode(), status=200, headers={
+    'Content-Type': '%s; charset=utf-8' % manifest_media_type,
+    'Docker-Content-Digest': manifest_digest,
   })
 
 
-def _rewrite_schema_if_necessary(namespace_name, repo_name, tag_name, manifest, parsed):
+def _rewrite_schema_if_necessary(namespace_name, repo_name, tag_name, manifest):
   # As per the Docker protocol, if the manifest is not schema version 1 and the manifest's
   # media type is not in the Accept header, we return a schema 1 version of the manifest for
   # the amd64+linux platform, if any, or None if none.
   # See: https://docs.docker.com/registry/spec/manifest-v2-2
   mimetypes = [mimetype for mimetype, _ in request.accept_mimetypes]
-  if parsed.media_type in mimetypes:
-    return parsed
+  if manifest.media_type in mimetypes:
+    return manifest.internal_manifest_bytes, manifest.digest, manifest.media_type
 
+  logger.debug('Manifest `%s` not compatible against %s; checking for conversion', manifest.digest,
+               request.accept_mimetypes)
   converted = registry_model.convert_manifest(manifest, namespace_name, repo_name, tag_name,
                                               mimetypes, storage)
   if converted is not None:
-    return converted
+    return converted.bytes, converted.digest, converted.media_type
 
   # For back-compat, we always default to schema 1 if the manifest could not be converted.
-  return registry_model.get_schema1_parsed_manifest(manifest, namespace_name, repo_name, tag_name,
-                                                    storage)
+  schema1 = registry_model.get_schema1_parsed_manifest(manifest, namespace_name, repo_name,
+                                                       tag_name, storage)
+  if schema1 is None:
+    return None, None, None
+
+  return schema1.bytes, schema1.digest, schema1.media_type
 
 
 def _reject_manifest2_schema2(func):