urllib.parse.unquote may return bytes

[pjw91]

In Py3, unquote does not accept bytes: the first line if '%' not in string would throw exception.

However, in Py2, that line will not throw exception and perform the checking.

When there's no %, the return value is the argument. That is, when the argument type is bytes, the return value is bytes, too.

Code snippet excerption from src/future/backports/urllib/parse.py#L515:

def unquote(string):
  if '%' not in string:
    string.split
    return string      # bytes or unicode, depends on argument; Py3 never reach here
  ...
  return ''.join(res)  # unicode

[jmadler]

@edschofield do we have any design decisions around when we wish to coerce and when not?

I'm inclined to suggest we align as much as possible with the Py3 behavior to make these proper polyfills, and instead provide tools for folks to coerce if they need at the call site, but that may not be practical.