Currently the error raised when bytes are passed to json.loads() is not very clear:
>>> json.loads(b'')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/wolf/dev/py/py3k/Lib/json/__init__.py", line 316, in loads
    return _default_decoder.decode(s)
  File "/home/wolf/dev/py/py3k/Lib/json/decoder.py", line 344, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
TypeError: can't use a string pattern on a bytes-like object

The attached patch changes the error message to:
>>> json.loads(b'')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/wolf/dev/py/py3k/Lib/json/__init__.py", line 315, in loads
    s.__class__.__name__))
TypeError: the JSON object must be str, not 'bytes'

(This came up on #18958, and it's required in order to check for a UTF-8 BOM without producing an even more misleading error message.)

Looks good to me: +1 :)

LGTM. But perhaps it's worth to test json.load() with binary file. I don't see much sense in testing json.loads() with numbers.

New changeset 241996bd3c6b by Ezio Melotti in branch 'default':
#19307: Improve error message for json.load(s) while passing objects of the wrong type.
http://hg.python.org/cpython/rev/241996bd3c6b

I added a test with json.load() too.  json.load(fp) just calls json.loads(fp.read()), but it's better to avoid white-box testing.