From 69262282fe4cd754f3fae5ad0202fbb321102b40 Mon Sep 17 00:00:00 2001
From: Joseph Schorr <josephschorr@users.noreply.github.com>
Date: Mon, 15 Feb 2016 10:57:20 -0500
Subject: [PATCH] Make sure to encode all V1 metadata strings

Fixes #1239
---
 endpoints/v2/manifest.py |  3 ++-
 test/registry_tests.py   | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 34 insertions(+), 1 deletion(-)

diff --git a/endpoints/v2/manifest.py b/endpoints/v2/manifest.py
index df116a87d..8840972d1 100644
--- a/endpoints/v2/manifest.py
+++ b/endpoints/v2/manifest.py
@@ -408,7 +408,8 @@ def _write_manifest(namespace, repo_name, manifest):
          images_map[v1_mdata.docker_id].storage.content_checksum != digest_str) or
         has_rewritten_ids):
 
-      working_docker_id = hashlib.sha256(mdata.v1_metadata_str + '@' + digest_str).hexdigest()
+      v1_metadata_str = mdata.v1_metadata_str.encode('utf-8')
+      working_docker_id = hashlib.sha256(v1_metadata_str + '@' + digest_str).hexdigest()
       logger.debug('Rewriting docker_id %s/%s %s -> %s', namespace, repo_name, v1_mdata.docker_id,
                    working_docker_id)
       has_rewritten_ids = True
diff --git a/test/registry_tests.py b/test/registry_tests.py
index d2507405a..f9cec9e2e 100644
--- a/test/registry_tests.py
+++ b/test/registry_tests.py
@@ -749,6 +749,38 @@ class RegistryTestsMixin(object):
     # Pull the repository.
     self.do_pull('public', 'newrepo', 'public', 'password', images=images, munge_shas=True)
 
+
+  def test_push_same_ids_different_sha_with_unicode(self):
+    if self.push_version == 'v1':
+      # No SHAs to munge in V1.
+      return
+
+    images = [
+      {
+        'id': 'latestid',
+        'contents': 'The latest image',
+        'unicode': u'the Pawe\xc5\x82 Kami\xc5\x84ski image',
+        'parent': 'baseid',
+      },
+      {
+        'id': 'baseid',
+        'contents': 'The base image',
+      }
+    ]
+
+    # Push a new repository.
+    self.do_push('public', 'newrepo', 'public', 'password', images=images)
+
+    # Pull the repository.
+    self.do_pull('public', 'newrepo', 'public', 'password', images=images)
+
+    # Push a the repository again, but with different SHAs.
+    self.do_push('public', 'newrepo', 'public', 'password', images=images, munge_shas=True)
+
+    # Pull the repository.
+    self.do_pull('public', 'newrepo', 'public', 'password', images=images, munge_shas=True)
+
+
   def test_push_pull_logging(self):
     # Push a new repository.
     self.do_push('public', 'newrepo', 'public', 'password')