Multipass yajl.dumps() breaks with unicode

[indigoviolet]

Summary is that if you have a unicode string as a value in a dict, and for some reason you want to do dumps({key: dumps(value)}), and then retrieve value, it breaks with py-yajl and not with standard library json. Is this something you're able to fix?

See console session below:

In [9]: u Out[9]: u'f\xe9in'

In [10]: print u -------> print(u) féin

In [11]: out = json.dumps({'a': json.dumps(u)})

In [12]: out Out[12]: '{"a": "\"f\u00e9in\""}'

In [13]: json.loads((json.loads(out))['a']) Out[13]: u'f\xe9in'

In [14]: out = yajl.dumps({'a': yajl.dumps(u)})

In [15]: out Out[15]: '{"a":"\"f\xc3\xa9in\""}'

In [16]: yajl.loads((yajl.loads(out))['a'])

UnicodeEncodeError Traceback (most recent call last)

in () UnicodeEncodeError: 'ascii' codec can't encode character u'\xe9' in position 2: ordinal not in range(128)

[indigoviolet]

Adding clearer console session

[indigoviolet]

Hey,

Do you have any idea when you're going to be able to get to this and #11?

Thanks!

[indigoviolet]

Argh. I closed this by mistake.

[rtyler]

I am pretty sure I resolved this in the latest commits, perhaps not.

FWIW you can join #yajl on Freenode if you have direct questions

[rtyler]

This should be resolved with the merging of SHA: 01cb3f10a3189100a91b6bdc2f5295f8f4782bfe