Ridiculous, broken, hacky and flawed implementation of StringPath, instances of which are both instances of pathlib.Path and of str

demo.txt

>>> p = StringPath("foo/bar/baz")
>>> p
StringPath(PosixPath('foo/bar/baz'))
>>> isinstance(p, pathlib.Path)
True
>>> isinstance(p, str)
True
>>> str(p)
'foo/bar/baz'
>>> p + 'hello'
'foo/bar/bazhello'
>>> p / 'hello'
PosixPath('foo/bar/baz/hello')
>>> 'hello' / p
PosixPath('hello/foo/bar/baz')
>>> p.split('r')
['foo/ba', '/baz']
>>> pathlib.Path(p)
PosixPath('foo/bar/baz')

stringpathhack.py

# A ridiculous, broken, hacky and flawed implementation of StringPath, 
# instances of which are both instances of pathlib.Path and of str
#
# ONLY FOR EXPERIMENTATION AND DEMONSTRATION PURPOSES
#
# Because pathlib.Path has __slots__ defined, it is not possible to
# directly subclass both Path and str. Hence, the terrible hack.
# "Tested" in Python 3.5

import pathlib

class StringPath(str):
    __class__ = pathlib.Path
    def __init__(self, path):
        if isinstance(path, str):
            path = pathlib.Path(path)
        self._path = path
        self.__dict__.update(
            { n : getattr(path, n) 
                 for n in dir(path) 
                     if not n.startswith('_')}
        )
        self._parts = path.parts
    
    def __repr__(self):
        return "StringPath({})".format(repr(self._path))
    
    def __truediv__(self, other):
        return self._path / other
    
    def __rtruediv__(self, other):
        return other / self._path