Merge pull request #541 from coreos-inc/python-registry-v2-yaml-sucks

Remove yaml and switch to JSON because yaml is so slow

endpoints/v2/manifest.py

@@ -4,7 +4,6 @@
 import logging
 import re
 import jwt.utils
-import yaml
 import json
 
 from flask import make_response, request, url_for
@@ -64,9 +63,7 @@ class SignedManifest(object):
   def __init__(self, manifest_bytes):
     self._bytes = manifest_bytes
 
-    # TODO(jakedt): If the manifest_bytes doesn't parse as valid YAML, safe_load returns the
-    # same string again. We should throw some sort of exception.
-    self._parsed = yaml.safe_load(manifest_bytes)
+    self._parsed = json.loads(manifest_bytes)
     self._signatures = self._parsed[_SIGNATURES_KEY]
     self._namespace, self._repo_name = self._parsed[_REPO_NAME_KEY].split('/')
     self._tag = self._parsed[_REPO_TAG_KEY]
@@ -110,9 +107,7 @@ class SignedManifest(object):
       image_digest = digest_tools.Digest.parse_digest(blob_sum_obj[_BLOB_SUM_KEY])
       metadata_string = history_obj[_V1_COMPAT_KEY]
 
-      # TODO(jakedt): If the metadata_string doesn't parse as valid YAML, safe_load returns the
-      # same string again. We should throw some sort of exception.
-      v1_metadata = yaml.safe_load(metadata_string)
+      v1_metadata = json.loads(metadata_string)
       command_list = v1_metadata.get('container_config', {}).get('Cmd', None)
       command = json.dumps(command_list) if command_list else None
 
@@ -123,12 +118,15 @@ class SignedManifest(object):
 
   @property
   def payload(self):
-    protected = self._signatures[0][_PROTECTED_KEY]
-    parsed_protected = yaml.safe_load(jwt.utils.base64url_decode(protected))
+    protected = str(self._signatures[0][_PROTECTED_KEY])
+
+    parsed_protected = json.loads(jwt.utils.base64url_decode(protected))
     logger.debug('parsed_protected: %s', parsed_protected)
+
     signed_content_head = self._bytes[:parsed_protected[_FORMAT_LENGTH_KEY]]
     logger.debug('signed content head: %s', signed_content_head)
-    signed_content_tail = jwt.utils.base64url_decode(parsed_protected[_FORMAT_TAIL_KEY])
+
+    signed_content_tail = jwt.utils.base64url_decode(str(parsed_protected[_FORMAT_TAIL_KEY]))
     logger.debug('signed content tail: %s', signed_content_tail)
     return signed_content_head + signed_content_tail
 

test/registry_tests.py

@@ -24,6 +24,7 @@ import endpoints.decorated
 import json
 import features
 import hashlib
+import logging
 
 import tarfile
 import shutil
@@ -48,6 +49,8 @@ except ValueError:
 # DB connection.
 
 testbp = Blueprint('testbp', __name__)
+logger = logging.getLogger(__name__)
+
 
 @testbp.route('/csrf', methods=['GET'])
 def generate_csrf():
@@ -101,6 +104,7 @@ class TestFeature(object):
 
 _PORT_NUMBER = 5001
 _CLEAN_DATABASE_PATH = None
+_JWK = RSAKey(key=RSA.generate(2048))
 
 
 def get_new_database_uri():
@@ -334,6 +338,9 @@ class V2RegistryPushMixin(V2RegistryMixin):
       checksum = 'sha256:' + hashlib.sha256(full_contents).hexdigest()
       builder.add_layer(checksum, json.dumps({'id': image_id, 'data': contents}))
 
+    # Build the manifest.
+    manifest = builder.build(_JWK)
+
     # Push the image's layers.
     for image_id, contents in images.iteritems():
       chunks = None
@@ -368,10 +375,6 @@ class V2RegistryPushMixin(V2RegistryMixin):
       self.conduct('PUT', location, params=dict(digest=checksum), expected_code=201, auth='jwt')
 
     # Write the manifest.
-    new_key = RSA.generate(2048)
-    jwk = RSAKey(key=new_key)
-    manifest = builder.build(jwk)
-
     self.conduct('PUT', '/v2/%s/%s/manifests/%s' % (namespace, repository, tag_name),
                  data=manifest.bytes, expected_code=202,
                  headers={'Content-Type': 'application/json'}, auth='jwt')