[tbworld]

To do the job right the 'post' parser would have to be rewritten with that feature in mind. The best you can do now is a hack that comes close -- depending on the length of the post and what 'bbcodes' and other non-text insertions have been made. Between the many exceptions that you have to take into account with the current parser and the extra code overhead you would be adding to the editor, it makes the project unfeasible in my opinion.
I