As noted in _loader._MemObjectProxy.to_json the way values are stringified is not compatible with json encoding, resulting in the following test failures:
======================================================================
FAIL: test_to_json (meliae.tests.test_loader.TestMemObj)
----------------------------------------------------------------------
Traceback (most recent call last):
File "meliae\tests\test_loader.py", line 239, in test_to_json
self.assertEqual(expected, [obj.to_json() for obj in objs])
AssertionError: Lists differ: ['{"address": 1, "type": "tupl... != ['{"address": 1, "type": "tupl...
First differing element 5:
{"address": 6, "type": "str", "size": 29, "value": "a str", "refs": []}
{"address": 6, "type": "str", "size": 29, "value": "a str"", "refs": []}
Diff is 720 characters long. Set self.maxDiff to None to see it.
======================================================================
FAIL: test_expand_refs_as_dict (meliae.tests.test_loader.TestObjManager)
----------------------------------------------------------------------
Traceback (most recent call last):
File "meliae\tests\test_loader.py", line 404, in test_expand_refs_as_dict
self.assertEqual({1: 'c', 'b': 'c'}, as_dict)
AssertionError: {1: 'c', 'b': 'c'} != {1: 'c"', 'b"': 'c"'}
- {1: 'c', 'b': 'c'}
+ {1: 'c"', 'b"': 'c"'}
? + + +
It would be possible to use json.encoder.encode_basestring_ascii to correct this, but currently json is not a hard dependency which complicates things a little.
This patch improves it so that the tests at least pass. Things may still break on some input data, though, so it might be better to use a real parser as bug 871451 suggests.