From 1f6ce63dde9d6601868d2c0fff6309fe20799917 Mon Sep 17 00:00:00 2001
From: Richard Maw <richard.maw@codethink.co.uk>
Date: Thu, 2 Oct 2014 11:26:01 +0000
Subject: Allow distbuild morphologies with binary data embedded

The horrible json.dumped, base64-encoded yaml dump is because we'd need
to rework the entire communications channel to not be json otherwise,
and we may want to do something else entirely with it, so to avoid
creating extra work for ourselves, let's just encapsulate non-utf8-clean
data.
---
 distbuild/serialise.py | 20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/distbuild/serialise.py b/distbuild/serialise.py
index 2b39000e..52ebc9c6 100644
--- a/distbuild/serialise.py
+++ b/distbuild/serialise.py
@@ -16,7 +16,9 @@
 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA..
 
 
+import base64
 import json
+import yaml
 
 import morphlib
 import logging
@@ -39,16 +41,14 @@ def serialise_artifact(artifact):
             'original_ref': source.original_ref,
             'sha1': source.sha1,
             'tree': source.tree,
-            'morphology': str(id(source.morphology)),
+            'morphology': id(source.morphology),
             'filename': source.filename,
 
-            # dict keys are converted to strings by json
-            # so we encode the artifact ids as strings
-            'artifact_ids': [str(id(artifact)) for (_, artifact)
+            'artifact_ids': [id(artifact) for (_, artifact)
                 in source.artifacts.iteritems()],
             'cache_id': source.cache_id,
             'cache_key': source.cache_key,
-            'dependencies': [str(id(d))
+            'dependencies': [id(d)
                 for d in source.dependencies],
         }
 
@@ -85,15 +85,17 @@ def serialise_artifact(artifact):
         if id(a) not in encoded_artifacts: # pragma: no cover
             encoded_artifacts[id(a)] = encode_artifact(a)
 
-    return json.dumps({'sources': encoded_sources,
+    content = {
+        'sources': encoded_sources,
         'artifacts': encoded_artifacts,
         'morphologies': encoded_morphologies,
-        'root_artifact': str(id(artifact)),
+        'root_artifact': id(artifact),
         'default_split_rules': {
             'chunk': morphlib.artifactsplitrule.DEFAULT_CHUNK_RULES,
             'stratum': morphlib.artifactsplitrule.DEFAULT_STRATUM_RULES,
         },
-    })
+    }
+    return json.dumps(base64.standard_b64encode(yaml.dump(content)))
 
 
 def deserialise_artifact(encoded):
@@ -148,7 +150,7 @@ def deserialise_artifact(encoded):
 
         return artifact
 
-    le_dicts = json.loads(encoded)
+    le_dicts = yaml.load(base64.standard_b64decode(json.loads(encoded)))
     artifacts_dict = le_dicts['artifacts']
     sources_dict = le_dicts['sources']
     morphologies_dict = le_dicts['morphologies']
-- 
cgit v1.2.1