Fix remaining unit tests about hyperjson

[mre]

Thanks to the help of the awesome @althonos in PyO3/pyo3#185, I managed to finally add bytes support. With that, we now have 130 successful and 31 skipped tests and the build is passing for the first time. Needless to say, this makes me very happy. 😆 🎉
Closing this.

[mre]

See #6

[mre]

Four tests are failing because it doesn't find JSONDecodeError. I tried to fix that, but to no avail.
Seems like I don't understand how exceptions get defined. 😅
Opened an issue here: PyO3/pyo3#171

[wdv4758h]

Multiple test cases are failing because serde_json::to_string output a compact result, while the std json output a extra space between key and value.

[mre]

@wdv4758h, good point.

I'm not sure whether we should treat that as a bug or an implementation detail. After all, JSON is a whitespace-agnostic format, so technically it should not make a difference. It might also cost us some performance to fix that in hyperjson and it doesn't seem like there will be an upstream change any time soon (see e.g. serde-rs/json#393).
So I suggest to keep that for a while and fix the affected unit tests by ignoring whitespace like so https://github.com/mre/hyperjson/blob/master/tests/test_dumps.py#L34. Sounds good?

[mre]

So we are down to four failing unit tests after fixing issues with JSONDecodeError. 🎉
The remaining tests are harder to fix, I guess and have to do with serde_json's (or Rusts) internal representation of types as strings.

For example:

self = <test_ujson.UltraJSONTests testMethod=test_encodeLongUnsignedConversion>

    def test_encodeLongUnsignedConversion(self):
        input = 18446744073709551615
        output = hyperjson.dumps(input)
        print(output)

>       self.assertEqual(input, hyperjson.loads(output))
E       AssertionError: 18446744073709551615 != 1.8446744073709552e+19

I don't know how to fix that yet. @wdv4758h, if you have an idea I'd be more than happy to listen. 😄

[mre]

Correction, we are down to four failing test now. Fixed the issues regarding long int parsing.
See e47ef32.

[mre]

Whoops, broke parse_int test with the last commit.
Should be fixed now. So we're down to three failing tests now, which all have to do with utf8 handling.
Python handles utf8 as unicode code points when serializing, I guess. Don't know what to do in this case. We will see.

[wdv4758h]

awesome progress !!! 🎆

[mre]

With 7a73455 I decided to ignore the unicode encoding test failure.
UTF-8 encoding/decoding works interchangeably with json and hyperjson, the only thing that does not work (and does not make much sense imho) is to check the string lengths.
Therefore I commented that part and just kept the rest.

One more test to go!