Fixes to ensuring existing code can process schema 2 manifests

endpoints/verbs/__init__.py

@@ -16,6 +16,7 @@ from data.registry_model import registry_model
 from endpoints.decorators import anon_protect, anon_allowed, route_show_if, parse_repository_name
 from endpoints.v2.blob import BLOB_DIGEST_ROUTE
 from image.appc import AppCImageFormatter
+from image.docker import ManifestException
 from image.docker.squashed import SquashedDockerImageFormatter
 from storage import Storage
 from util.audit import track_and_log, wrap_repository
@@ -42,7 +43,7 @@ class VerbReporter(TarLayerFormatterReporter):
     metric_queue.verb_action_passes.Inc(labelvalues=[self.kind, pass_count])
 
 
-def _open_stream(formatter, tag, manifest, derived_image_id, handlers, reporter):
+def _open_stream(formatter, tag, manifest, schema1_manifest, derived_image_id, handlers, reporter):
   """
   This method generates a stream of data which will be replicated and read from the queue files.
   This method runs in a separate process.
@@ -68,7 +69,7 @@ def _open_stream(formatter, tag, manifest, derived_image_id, handlers, reporter)
     for layer in reversed(layers):
       yield image_stream_getter(store, layer.blob)
 
-  stream = formatter.build_stream(tag, manifest, derived_image_id, layers,
+  stream = formatter.build_stream(tag, schema1_manifest, derived_image_id, layers,
                                   tar_stream_getter_iterator, reporter=reporter)
 
   for handler_fn in handlers:
@@ -220,9 +221,21 @@ def _verify_repo_verb(_, namespace, repo_name, tag_name, verb, checker=None):
     logger.debug('Could not get manifest on %s/%s:%s::%s', namespace, repo_name, tag.name, verb)
     abort(404)
 
+  # Ensure the manifest is not a list.
+  try:
+    schema1_manifest = registry_model.get_schema1_parsed_manifest(manifest, namespace,
+                                                                  repo_name, tag.name,
+                                                                  storage)
+  except ManifestException:
+    logger.exception('Could not get manifest on %s/%s:%s::%s', namespace, repo_name, tag.name, verb)
+    abort(400)
+
+  if schema1_manifest is None:
+    abort(404)
+
   # If there is a data checker, call it first.
   if checker is not None:
-    if not checker(tag, manifest):
+    if not checker(tag, schema1_manifest):
       logger.debug('Check mismatch on %s/%s:%s, verb %s', namespace, repo_name, tag.name, verb)
       abort(404)
 
@@ -230,12 +243,12 @@ def _verify_repo_verb(_, namespace, repo_name, tag_name, verb, checker=None):
   assert tag.repository.namespace_name
   assert tag.repository.name
 
-  return tag, manifest
+  return tag, manifest, schema1_manifest
 
 
 def _repo_verb_signature(namespace, repository, tag_name, verb, checker=None, **kwargs):
   # Verify that the tag exists and that we have access to it.
-  tag, manifest = _verify_repo_verb(storage, namespace, repository, tag_name, verb, checker)
+  tag, manifest, _ = _verify_repo_verb(storage, namespace, repository, tag_name, verb, checker)
 
   # Find the derived image storage for the verb.
   derived_image = registry_model.lookup_derived_image(manifest, verb,
@@ -261,7 +274,8 @@ def _repo_verb(namespace, repository, tag_name, verb, formatter, sign=False, che
   # Verify that the image exists and that we have access to it.
   logger.debug('Verifying repo verb %s for repository %s/%s with user %s with mimetype %s',
                verb, namespace, repository, get_authenticated_user(), request.accept_mimetypes.best)
-  tag, manifest = _verify_repo_verb(storage, namespace, repository, tag_name, verb, checker)
+  tag, manifest, schema1_manifest = _verify_repo_verb(storage, namespace, repository,
+                                                      tag_name, verb, checker)
 
   # Load the repository for later.
   repo = model.repository.get_repository(namespace, repository)
@@ -323,7 +337,7 @@ def _repo_verb(namespace, repository, tag_name, verb, formatter, sign=False, che
   # and send the results to the client and storage.
   handlers = [hasher.update]
   reporter = VerbReporter(verb)
-  args = (formatter, tag, manifest, derived_image.unique_id, handlers, reporter)
+  args = (formatter, tag, manifest, schema1_manifest, derived_image.unique_id, handlers, reporter)
   queue_process = QueueProcess(
     _open_stream,
     8 * 1024,
@@ -360,7 +374,7 @@ def _repo_verb(namespace, repository, tag_name, verb, formatter, sign=False, che
 def os_arch_checker(os, arch):
   def checker(tag, manifest):
     try:
-      image_json = json.loads(manifest.get_parsed_manifest().leaf_layer.raw_v1_metadata)
+      image_json = json.loads(manifest.leaf_layer.raw_v1_metadata)
     except ValueError:
       logger.exception('Could not parse leaf layer JSON for manifest %s', manifest)
       return False