What is the difference between json.dumps/loads and tornado.escape.json_encode/json_decode?

Question

It seems they are behaving exactly the same way.

>>> data
[('a', 'b'), {'a': 1, 'b': 2}, ['a', 'b'], 'a', 'b']
>>> json.dumps(data)
'[["a", "b"], {"a": 1, "b": 2}, ["a", "b"], "a", "b"]'
>>> tornado.escape.json_encode(data)
'[["a", "b"], {"a": 1, "b": 2}, ["a", "b"], "a", "b"]'
>>> json.loads(json.dumps(data))
[[u'a', u'b'], {u'a': 1, u'b': 2}, [u'a', u'b'], u'a', u'b']
>>> tornado.escape.json_decode(json.dumps(data))
[[u'a', u'b'], {u'a': 1, u'b': 2}, [u'a', u'b'], u'a', u'b']

Answer 1

Sometimes it's useful to read the source code:

def json_encode(value):
    return json.dumps(value).replace("</", "<\\/")

def json_decode(value):
    return json.loads(to_basestring(value))

def to_basestring(value):
    if isinstance(value, _BASESTRING_TYPES):
        return value
    assert isinstance(value, bytes_type)
    return value.decode("utf-8")

to_basestring is mostly needed for python 3.x to ensure the value has type str, not bytes, because json.loads can't deal with the latter.