test_invalid_utf8: Use distinct if/else branches

[jamessan]

pytest isn't able to report the expected value when using the X if Y else Z syntax.

See pytest-dev/pytest#7727

[jamessan]

Forcing an error does show the expected reporting

===================================================================================================== FAILURES =====================================================================================================
________________________________________________________________________________________________ test_invalid_utf8 _________________________________________________________________________________________________

vim = <pynvim.api.nvim.Nvim object at 0xb608d1f0>

    def test_invalid_utf8(vim):
        vim.command('normal "=printf("%c", 0xFF)\np')
        assert vim.eval("char2nr(getline(1))") == 0xFF
        if IS_PYTHON3:
            assert vim.current.buffer[:] == ['\udcff']
        else:
            assert vim.current.buffer[:] == ['\xff']
    
        vim.current.line += 'x'
        if IS_PYTHON3:
            assert vim.eval("getline(1)", decode=False) == '\udcffx'
>           assert vim.current.buffer[:] == ['\udcffxy']
E           AssertionError: assert ['\udcffx'] == ['\udcffxy']
E             At index 0 diff: '\udcffx' != '\udcffxy'
E             Full diff:
E             - ['\udcffx']
E             + ['\udcffxy']
E             ?          +

test/test_buffer.py:175: AssertionError

[jamessan]

Guess I didn't need to force a failure to see how this looks. Python 2 tests are failing and probably have been since they were added:

    def test_invalid_utf8(vim):
        vim.command('normal "=printf("%c", 0xFF)\np')
        assert vim.eval("char2nr(getline(1))") == 0xFF
    
        if IS_PYTHON3:
            assert vim.current.buffer[:] == ['\udcff']
        else:
>           assert vim.current.buffer[:] == ['\xff']
E           AssertionError: assert [''] == ['\xff']
E             At index 0 diff: u'' != '\xff'
E             Full diff:
E             - [u'']
E             + ['\xff']

[blueyed]

Note that assert 1 == 2 if 0 else 3 does not fail, but only assert 1 == (2 if 0 else 3)..!?

[blueyed]

Distinct branches are good for code coverage though anyway.

[jamessan]

I'm not sure what the correct behavior is supposed to be for Py2. @Shougo or @bfredl, can you advise?

[bfredl]

the correct default behaviour in python2 should be no decoding, i e get the raw '\xFF' byte out (and no unicode string)

[jamessan]

the correct default behaviour in python2 should be no decoding

Maybe you can push the right changes to my branch? I'm not familiar with the pynvim APIs.

[jamessan]

Closing in favor of #492, which rips out the py2 code, thus getting rid of the bug.