replace all but alphanumeric and some punctuation

[davedpss]

I am trying to clean up some text that has some white space that I can't seem to get rid of. I have tried replacing \s, \t,\n, \r, \r\n etc. and that strips out most of the new lines but not all. When I look at in a text editor such as TextWrangler, I am seeing an upside down question mark where these spaces are. I thought instead of trying to match something and then deleting that I could try to delete everything but what matched. I don't know how to do this with regular expressions. If anybody had a pattern and a replacement that would work, I would appreciate it.

Thanks - Dave.

[diafol]

Anything to do with encoding?