LGTM.com errors

[DimitriPapadopoulos]

Fix part of these LGTM.com errors:
https://lgtm.com/projects/g/nltk/nltk/?severity=error

[dannysepler]

should it say "invalid" instead?

[DimitriPapadopoulos]

The ReadError documentation reads:

    :param expected: What was expected when an error occurred.

[dannysepler]

i see. i guess i am used to errors telling me what's wrong, rather than what is expected. but honestly, i don't even think this code works!! from my own poking around, eval-ing a string with an invalid escape sequence does NOT raise a ValueError, it instead raises a DeprecationWarning. This code dates back to 2008, so perhaps then python used raise an error?

There seems to be active discussion in the python community about this, and they may raise an error down the road.

all of this is to say, i know you're just trying to improve small bits of code quality and not anything more. So I added a quick unit test to check other errors work (they seem to), and this looks fine now!

[stevenbird]

I think the error message needs to make it easy for people to fix. Doesn't that mean reporting the invalid string?

[DimitriPapadopoulos]

Perhaps, but then that's because the ReadError class is poorly designed. This could be fixed in a different PR.

[purificant]

ReadError class is defined further up in the file as:

class ReadError(ValueError):
    """
    Exception raised by read_* functions when they fail.
    :param position: The index in the input string where an error occurred.
    :param expected: What was expected when an error occurred.
    """

    def __init__(self, expected, position):
        ValueError.__init__(self, expected, position)
        self.expected = expected
        self.position = position

    def __str__(self):
        return f"Expected {self.expected} at {self.position}"

This PR fixed a bug where constructor was only passed one argument instead of two and the new string displayed to the user in the context of line 245 will be f"Expected valid escape sequence at {start_position}".

[dannysepler]

Thanks, @DimitriPapadopoulos !