Fix support for pulling manifest lists via Docker V1 protocol where applicable

data/registry_model/interface.py

@@ -12,6 +12,12 @@ class RegistryDataInterface(object):
     """ Returns whether the implementation of the data interface supports schema 2 format
         manifests. """
 
+  @abstractmethod
+  def get_legacy_tags_map(self, repository_ref, storage):
+    """ Returns a map from tag name to its legacy image ID, for all tags with legacy images in
+        the repository. Note that this can be a *very* heavy operation.
+    """
+
   @abstractmethod
   def find_matching_tag(self, repository_ref, tag_names):
     """ Finds an alive tag in the repository matching one of the given tag names and returns it

data/registry_model/registry_oci_model.py

@@ -13,6 +13,7 @@ from data.registry_model.datatypes import (Tag, Manifest, LegacyImage, Label, Se
 from data.registry_model.shared import SharedModel
 from data.registry_model.label_handlers import apply_label_to_manifest
 from image.docker import ManifestException
+from image.docker.schema2 import DOCKER_SCHEMA2_MANIFESTLIST_CONTENT_TYPE
 
 
 logger = logging.getLogger(__name__)
@@ -28,6 +29,31 @@ class OCIModel(SharedModel, RegistryDataInterface):
         manifests. """
     return True
 
+  def get_legacy_tags_map(self, repository_ref, storage):
+    """ Returns a map from tag name to its legacy image ID, for all tags with legacy images in
+        the repository. Note that this can be a *very* heavy operation.
+    """
+    tags = oci.tag.list_alive_tags(repository_ref._db_id)
+    legacy_images_map = oci.tag.get_legacy_images_for_tags(tags)
+
+    tags_map = {}
+    for tag in tags:
+      legacy_image = legacy_images_map.get(tag.id)
+      if legacy_image is not None:
+        tags_map[tag.name] = legacy_image.docker_image_id
+      else:
+        manifest = Manifest.for_manifest(tag.manifest, None)
+        if legacy_image is None and manifest.media_type == DOCKER_SCHEMA2_MANIFESTLIST_CONTENT_TYPE:
+          # See if we can lookup a schema1 legacy image.
+          v1_compatible = self.get_schema1_parsed_manifest(manifest, '', '', '', storage)
+          if v1_compatible is not None:
+            v1_id = v1_compatible.leaf_layer_v1_image_id
+            if v1_id is not None:
+              tags_map[tag.name] = v1_id
+
+
+    return tags_map
+
   def find_matching_tag(self, repository_ref, tag_names):
     """ Finds an alive tag in the repository matching one of the given tag names and returns it
         or None if none.

data/registry_model/registry_pre_oci_model.py

@@ -32,6 +32,13 @@ class PreOCIModel(SharedModel, RegistryDataInterface):
         manifests. """
     return False
 
+  def get_legacy_tags_map(self, repository_ref, storage):
+    """ Returns a map from tag name to its legacy image, for all tags with legacy images in
+        the repository.
+    """
+    tags = self.list_repository_tags(repository_ref, include_legacy_images=True)
+    return {tag.name: tag.legacy_image.docker_image_id for tag in tags}
+
   def find_matching_tag(self, repository_ref, tag_names):
     """ Finds an alive tag in the repository matching one of the given tag names and returns it
         or None if none.

data/registry_model/test/test_interface.py

@@ -231,6 +231,8 @@ def test_repository_tags(repo_namespace, repo_name, registry_model):
   tags = registry_model.list_repository_tags(repository_ref, include_legacy_images=True)
   assert len(tags)
 
+  tags_map = registry_model.get_legacy_tags_map(repository_ref, storage)
+
   for tag in tags:
     found_tag = registry_model.get_repo_tag(repository_ref, tag.name, include_legacy_image=True)
     assert found_tag == tag
@@ -241,6 +243,8 @@ def test_repository_tags(repo_namespace, repo_name, registry_model):
     found_image = registry_model.get_legacy_image(repository_ref,
                                                   found_tag.legacy_image.docker_image_id)
     assert found_image == found_tag.legacy_image
+    assert tag.name in tags_map
+    assert tags_map[tag.name] == found_image.docker_image_id
 
 
 def test_repository_tag_history(registry_model):

endpoints/v1/tag.py

@@ -27,11 +27,7 @@ def get_tags(namespace_name, repo_name):
     if repository_ref is None:
       abort(404)
 
-    # TODO(jschorr): Change this to normalize manifest lists back to their legacy image
-    # (if applicable).
-    tags = registry_model.list_repository_tags(repository_ref, include_legacy_images=True)
-    tag_map = {tag.name: tag.legacy_image.docker_image_id
-               for tag in tags if tag.legacy_image_if_present}
+    tag_map = registry_model.get_legacy_tags_map(repository_ref, storage)
     return jsonify(tag_map)
 
   abort(403)

test/registry/registry_tests.py

@@ -1368,7 +1368,7 @@ def test_push_tag_existing_image(v1_protocol, puller, basic_images, liveserver_s
   True,
   False
 ])
-def test_push_pull_manifest_list_back_compat(v22_protocol, v2_protocol, basic_images,
+def test_push_pull_manifest_list_back_compat(v22_protocol, legacy_puller, basic_images,
                                              different_images, liveserver_session, app_reloader,
                                              schema_version, data_model, is_amd):
   """ Test: Push a new tag with a manifest list containing two manifests, one (possibly) legacy
@@ -1405,9 +1405,9 @@ def test_push_pull_manifest_list_back_compat(v22_protocol, v2_protocol, basic_im
 
   # Pull the tag and ensure we (don't) get back the basic images, since they are(n't) part of the
   # amd64+linux manifest.
-  v2_protocol.pull(liveserver_session, 'devtable', 'newrepo', 'latest', basic_images,
-                   credentials=credentials,
-                   expected_failure=Failures.UNKNOWN_TAG if not is_amd else None)
+  legacy_puller.pull(liveserver_session, 'devtable', 'newrepo', 'latest', basic_images,
+                     credentials=credentials,
+                     expected_failure=Failures.UNKNOWN_TAG if not is_amd else None)
 
 
 @pytest.mark.parametrize('schema_version', [