Make sure to decode manifests into utf-8 when necessary

This fixes a decoding error

image/docker/schema2/test/test_manifest.py

@@ -1,5 +1,8 @@
+# -*- coding: utf-8 -*-
+
 import json
 import pytest
+import os
 
 from app import docker_v2_signing_key
 from image.docker.schema1 import (DockerSchema1ManifestBuilder,
@@ -7,6 +10,7 @@ from image.docker.schema1 import (DockerSchema1ManifestBuilder,
                                   DOCKER_SCHEMA1_MANIFEST_CONTENT_TYPE)
 from image.docker.schema2.manifest import (MalformedSchema2Manifest, DockerSchema2Manifest,
                                            DockerSchema2ManifestBuilder, EMPTY_LAYER_BLOB_DIGEST)
+from image.docker.schema2.config import DockerSchema2Config
 from image.docker.schema2.test.test_config import CONFIG_BYTES
 from image.docker.schemautil import ContentRetrieverForTesting
 
@@ -351,3 +355,67 @@ def test_remote_layer_manifest():
 
   assert set(manifest.blob_digests) == {'sha256:adef', 'sha256:abcd', 'sha256:1352', 'sha256:1353'}
   assert set(manifest.local_blob_digests) == {'sha256:abcd', 'sha256:1352', 'sha256:1353'}
+
+
+def test_unencoded_unicode_manifest():
+  builder = DockerSchema2ManifestBuilder()
+  builder.add_layer('sha256:abc123', 123)
+  builder.set_config_digest('sha256:def456', 2000)
+  manifest = builder.build()
+
+  retriever = ContentRetrieverForTesting.for_config({
+    "config": {
+      "author": u"Sômé guy",
+    },
+    "rootfs": {"type": "layers", "diff_ids": []},
+    "history": [
+      {
+        "created": "2018-04-03T18:37:09.284840891Z",
+        "created_by": "base",
+        "author": u"Sômé guy",
+      },
+    ],
+  }, 'sha256:def456', 2000, ensure_ascii=False)
+
+  layers = list(manifest.get_layers(retriever))
+  assert layers[0].author == u"Sômé guy"
+
+
+def test_build_unencoded_unicode_manifest():
+  config_json = json.dumps({
+    "config": {
+      "author": u"Sômé guy",
+    },
+    "rootfs": {"type": "layers", "diff_ids": []},
+    "history": [
+      {
+        "created": "2018-04-03T18:37:09.284840891Z",
+        "created_by": "base",
+        "author": u"Sômé guy",
+      },
+    ],
+  }, ensure_ascii=False)
+
+  schema2_config = DockerSchema2Config(config_json)
+
+  builder = DockerSchema2ManifestBuilder()
+  builder.set_config(schema2_config)
+  builder.add_layer('sha256:abc123', 123)
+  builder.build()
+
+
+def test_load_unicode_manifest():
+  test_dir = os.path.dirname(os.path.abspath(__file__))
+  with open(os.path.join(test_dir, 'unicode_manifest_config.json'), 'r') as f:
+    retriever = ContentRetrieverForTesting()
+    retriever.add_digest('sha256:5bdd65cdd055c7f3bbaecdc9fd6c75f155322520f85953aa0e2724cab006d407',
+                         f.read())
+
+  with open(os.path.join(test_dir, 'unicode_manifest.json'), 'r') as f:
+    manifest_bytes = f.read()
+
+  manifest = DockerSchema2Manifest(manifest_bytes)
+  assert manifest.digest == 'sha256:97556fa8c553395bd9d8e19a04acef4716ca287ffbf6bde14dd9966053912613'
+
+  layers = list(manifest.get_layers(retriever))
+  assert layers[-1].author == u"Sômé guy"