If you are parsing a "PPT" file, I would approach that problem by writing a perl script that
If any character is outside of this range, then that character is filtered out. Also, if it finds the word ":the#$" by itself alone, then it skips that too since we're looking for at least 6 characters next to each other that fall within the expected range. This would be an easy way to filter out all the binary "trash" that ppt files are filled with. So, I'd start there. Of course, if it's a PPTX file, then you just unzip it and run some type of html or xml filter on the text, and you get the content that way. Easy! ;-)