Update the layer rename PR to preserve the original manifest

endpoints/v2/manifest.py

@@ -185,13 +185,6 @@ class SignedManifestBuilder(object):
       _V1_COMPAT_KEY: v1_json_metadata,
     })
 
-  def add_top_layer(self, layer_digest, v1_json_metadata):
-    self._fs_layer_digests.insert(0, {
-      _BLOB_SUM_KEY: layer_digest,
-    })
-    self._history.insert(0, {
-      _V1_COMPAT_KEY: v1_json_metadata,
-    })
 
   def build(self, json_web_key):
     """ Build the payload and sign it, returning a SignedManifest object.
@@ -395,53 +388,33 @@ def _write_manifest(namespace, repo_name, manifest):
   # a storage with a content checksum different from the existing, then we need to rewrite
   # the Docker ID to ensure consistency.
   tag_name = manifest.tag
-  updated_manifest_builder = SignedManifestBuilder(namespace, repo_name, tag_name)
+  has_rewritten_ids = False
-  has_updated_manifest = False
   updated_id_map = {}
 
   for mdata in layers:
     digest_str = str(mdata.digest)
     v1_mdata = mdata.v1_metadata
-    updated_id_map[v1_mdata.docker_id] = v1_mdata.docker_id
+    working_docker_id = v1_mdata.docker_id
 
     # Ensure that all blobs exist.
     blob_storage = storage_map.get(digest_str)
     if blob_storage is None:
       raise BlobUnknown(detail={'digest': digest_str})
 
-    if v1_mdata.docker_id in images_map:
+    # Ensure that the V1 image's storage matches the V2 blob. If not, we've found
-      # Ensure that the V1 image's storage matches the V2 blob. If not, we've found
+    # a data inconsistency and need to create a new layer ID for the V1 image, and all images
-      # a data inconsistency and need to create a new layer ID for the V1 image.
+    # that follow it in the ancestry chain.
-      v1_image = images_map[v1_mdata.docker_id]
+    if ((v1_mdata.docker_id in images_map and
-      if has_updated_manifest or v1_image.storage.content_checksum != digest_str:
+         images_map[v1_mdata.docker_id].storage.content_checksum != digest_str) or
-        new_synthetic_id = hashlib.sha256(mdata.v1_metadata_str + '@' + digest_str).hexdigest()
+        has_rewritten_ids):
-        logger.debug('Got content mismatch for layer %s under repo %s/%s. New ID: %s',
-                     v1_mdata.docker_id, namespace, repo_name, new_synthetic_id)
 
-        updated_id_map[v1_mdata.docker_id] = new_synthetic_id
+      working_docker_id = hashlib.sha256(mdata.v1_metadata_str + '@' + digest_str).hexdigest()
-        has_updated_manifest = True
+      logger.debug('Rewriting docker_id %s/%s %s -> %s', namespace, repo_name, v1_mdata.docker_id,
+                   working_docker_id)
+      has_rewritten_ids = True
 
-    # Update the manifest withn the new ID (if any).
+    # Store the new docker id in the map
-    v1_metadata_json = mdata.v1_metadata_str
+    updated_id_map[v1_mdata.docker_id] = working_docker_id
-    if has_updated_manifest:
-      v1_metadata_json = _updated_v1_metadata(mdata.v1_metadata_str, updated_id_map)
-
-    updated_manifest_builder.add_top_layer(digest_str, v1_metadata_json)
-
-  # If the manifest was changed due to an updated layer ID, then create a new manifest
-  # based on the updated data.
-  if has_updated_manifest:
-    manifest = updated_manifest_builder.build(docker_v2_signing_key)
-    layers = list(manifest.layers)
-
-  # Synthesize the V1 metadata for each layer.
-  for mdata in layers:
-    v1_mdata = mdata.v1_metadata
-
-    # If the layer with the V1 id already exists, then nothing more to do. We've ensured
-    # it points to the correct content SHA above.
-    if v1_mdata.docker_id in images_map:
-      continue
 
     # Lookup the parent image for the layer, if any.
     parent_image = None
@@ -454,9 +427,14 @@ def _write_manifest(namespace, repo_name, manifest):
     # Synthesize and store the v1 metadata in the db.
     digest_str = str(mdata.digest)
     blob_storage = storage_map[digest_str]
-    image = model.image.synthesize_v1_image(repo, blob_storage, v1_mdata.docker_id,
+
+    v1_metadata_json = mdata.v1_metadata_str
+    if has_rewritten_ids:
+      v1_metadata_json = _updated_v1_metadata(mdata.v1_metadata_str, updated_id_map)
+
+    image = model.image.synthesize_v1_image(repo, blob_storage, working_docker_id,
                                             v1_mdata.created, v1_mdata.comment, v1_mdata.command,
-                                            mdata.v1_metadata_str, parent_image)
+                                            v1_metadata_json, parent_image)
 
     images_map[v1_mdata.docker_id] = image
 
@@ -466,9 +444,9 @@ def _write_manifest(namespace, repo_name, manifest):
 
   # Store the manifest pointing to the tag.
   manifest_digest = manifest.digest
-  leaf_layer = layers[-1]
+  leaf_layer_id = images_map[layers[-1].v1_metadata.docker_id].docker_image_id
-  model.tag.store_tag_manifest(namespace, repo_name, tag_name, leaf_layer.v1_metadata.docker_id,
+  model.tag.store_tag_manifest(namespace, repo_name, tag_name, leaf_layer_id, manifest_digest,
-                               manifest_digest, manifest.bytes)
+                               manifest.bytes)
 
   # Spawn the repo_push event.
   event_data = {

test/registry_tests.py

@@ -678,7 +678,7 @@ class V2RegistryPullMixin(V2RegistryMixin):
       found_v1_layers.add(v1_history['id'])
 
     for image in images:
-      self.assertEquals(not munge_shas, image['id'] in found_v1_layers)
+      self.assertIn(image['id'], found_v1_layers)
 
     return blobs