JSON doesn't handle arbitrary bytes

fail.py

In [1]: arb = open('random-crap', 'rb').read()

In [2]: import json

In [3]: arb
Out[3]: '\x88\xb5rz\xe8(\xcd\x7f\xbc\x8e\xae\xfe\xfcA\xd7Q\xa8NOPQg\rZ\xe4\xb7\xf7p\xc4U<\xc6.\x94\xc8\xc3M\x91\xcdN\x91%j\xa8\xfd\xef\xcaI\xa1\xb2XB\xa5#\xae~\n\xc1\x8b0B\xe1^F.E\xdc\xea`<\xd7\x88&pR\xf0\xd8\xde\xabA\\\xc1\x10\x91\x8bsQ\x8a[\x15\x0c%G2\x85\x14\x97\t\xbe\x99T\xc8\x02\xac\xc6\xec\x1a\xe4\xa0x\x99]`\x90\xca&\x9a\x15\x08\x84W\x9eO\\\xa3NE\xd0YyA\xed|\xde\x1dl\xb8\xccU^\xb9\x83\xfe4\xd8\xa4\x0c2\xfb;\xff\xf3Qw\xeeX\xd6&\xef\x0f\xdc\xc7i\x93\xbd\xbav\xcb\x9a\x85%\x0c\xe2\x1dc\xb6\x8dcLN\x98\x9a?\x18{\x80`o\x0c\x07\xdf\xf9\x03\x01\xd6H\xa1\x11r\x16Gm\xdfW\xbd\xac\xb0!\x8f8\xa8\x99\n\x1c\x16(\x00\xdb~\xe3\x11\x8d?\xba\xc5\xc9\x94\x963(:\x99\xa8\x99\xc6\xae\x1f\xf1\xa2\x1c\xfc\x9b\x92\xb4\xb0\xf5!\x19\x01@\xb8\xb7\x8bOx\x19\x0fP\xfe\x86\xb7\x04\x05\xcd\xac\x8e@qC\xa7\xdb\x85\xae\x92\xa9:\x81_\xf1\x1b\xfd\xcb\x87\xb9\x88r\x94\xe6 \xfeG\xfe\xcd\xf2\x1d\x152\x04g\xd0\xe9O\xd0\xf5\x04\xc0\x9c\xb44\xa3\x84\x9f\xbd\xd0\xe3b\x90UaS\x00\xe4\xdd\xb2\x95\xcc\xe7\xd7l\xba\xdd\xb1\x1d\x81\xe4\x0e\xad\xee\x7fB+N\x18\xf5\x8c&\x0c\xf9\x94R\x86A\x1a\xbe\xdas\x98\x12*\x16\xfc\xba\xc7j\x10\x87E\xc8\xde\xc6\xea\x1d&\x17K6\x86\xc6\x99Y\x14"x\x06t\xd0\x89R\xafg\x8e\xca\x95BN\xcb\xa6\x84\x16\xaf\t\xd3\t\x7f\xe4\xfe\xf4>\xed\x93\xff\xed^\x00kd?\x8b~w\x05\xbe\xc4l$\xff\x1d\xb5\\#\x8d\x96\xa6\xde\xae1\x8f\x0e\x19D\xcf\xd2\x16\x17\xbe\x1c\xb5\xb1\xab\xc5\x07.\'\xb8\x01p\xbekUo\x1d\x8f\xecx\xaeK\xde\x9f\xa2\xffB\xe8\xdd\xd5r\x89WJ\xf6w\x17aG:WU^\xa1H,\x98\xd5\xc7\xf2\xd4\x1f\xf6\xc6\x0bi\xf6\x91,N6\xf3\xa2\x8b\xcc\xb3\xed\t5\xc0{\xcc\xe4\xf7\xe6\x02\x06\xe9.)\x90\x10G\xb69B\x9e\xd7\x14{\x8c\xe4\xaa\xf6\xfb\xf70\xef\xd3\xdd\x1a\xc1\xaa\x92\xad\x0cll\xf6\xd1`\xe4m4\xa8{\x896\x11\xd0:=\n~\x08S\x93Zu\xfb\x86\xe9\xed\xde\x1e5\x9d_\x8a\xd8\x8a\x03H\xd5\\\xc1\xab\x9a\xc7_\xbd[W\xc8\xb4\xe8=U\x8f\xe3\xf0\t\xccL^\xf0\xf3\xd3\xa2\xa5X)\x81S\x82\x9a\x10w1Qk,C\xfd\xa6\x0c\x91aej\x03\r\xcf\xbc\xd9\xc0\xba\x1b]\x03\xd3\x82P\xdd/\xb6\xf5v\xa7\x9c\xf0.\x10\xc0\x8d5\xaf\xfdv\x18a\xe9\x0e\xc4\xdd\xb0\x15\x90\x12\x98#\xf69U\xf1\x8e\x1a\xce\x95\\\x91\x02\xbe\x13\x88\xa6i\xe2\x1f7\xb3\xdb\x0c}\xc2\xe5\x97A\xb8\x90l\xad\x9f\xf7CD\xb4\x8f\x0b\xe9#X\x15\x8ei\x89\xafyQ)\xc0\xe1\x95\xe6/\xc5c\xe0\x14\x80H\xf7\xb1M\xbf\xe8\xbf\xd96\xac\xb6b@\xcd>\x19\xb6y4\xe5\xed\xb3<\x1a\xf0\xfb\xe1\xe1\xee3\xa8[c\xda\xd8\xa3CS\xa0\x86\x98KG,\x80o\xc0\x7f\xf3\xf0\xba"Zm\x95\xa9U#\xd2\xa3Q\xcac\xe7\x10\x8fk\xe2W\xb1;\xc5\x8eU\xaf\xd2G\x9f\xc0\t\x96\x1b\x08\xa8;\xf1y\xe1\x06\x1bi\x98\x82\xbd\x05\x93\xdbZ\xddIW{\xb8^\x1a\x92\x1bKlD\xe0\xa4\xd1\x14\xff\xb8L\\\x18\xcd\xc6\xe7\x8e\x93\xc5f\x0bz\x88\xd4f\xb2T\x17h\x1e\xe2VU\xc2\x88\xdf\t\xe2\xb0\xaes\x89V\xd0\xb4\xd6\x8c\x9a\x1e-7\\\xc2ly\xf9\xd1\xc1aj\x90\x9e\xb3\x9bLj\xa1\xe1n\xfd\x1b^\xd6Bf(\x16\xbb\x80\x85\x82k\xfa\x14\x88\xd3[\xdc\x95\xa3\x19rL\x92)\x0ba\x82Q)U|\x7fT\xa4R\x98\x93\xb3\xe3\xa7\xb8\xd0}@\xe1\xff\x07\xac\xff\xba\x99\x0ehk.\xed\r\'~\xc2\xac\x98r<\xbbUNs\xfb\xb5\xdc>\x1f\xde\xab\x0fq\xd2\xf6:\x87\xaf\xa5\xa7$\x03\xc2W\xa1'

In [4]: msg = json.dumps(arb)                                                                                                                                                                                                                                                         
---------------------------------------------------------------------------
UnicodeDecodeError                        Traceback (most recent call last)
<ipython-input-4-845ccbae793b> in <module>()
----> 1 msg = json.dumps(arb)

/usr/local/Cellar/python/2.7.2/lib/python2.7/json/__init__.pyc in dumps(obj, skipkeys, ensure_ascii, check_circular, allow_nan, cls, indent, separators, encoding, default, **kw)
    229         cls is None and indent is None and separators is None and
    230         encoding == 'utf-8' and default is None and not kw):
--> 231         return _default_encoder.encode(obj)
    232     if cls is None:
    233         cls = JSONEncoder

/usr/local/Cellar/python/2.7.2/lib/python2.7/json/encoder.pyc in encode(self, o)
    193                     o = o.decode(_encoding)
    194             if self.ensure_ascii:
--> 195                 return encode_basestring_ascii(o)
    196             else:
    197                 return encode_basestring(o)

UnicodeDecodeError: 'utf8' codec can't decode byte 0x88 in position 0: invalid start byte

gen.sh

dd if=/dev/urandom of=random-crap count=1000 bs=1