LPeg capturing too much?

[Peter Odding <xolox@...>]

Hi all! I’m trying to write an LPeg parser for semi-structured text ala Textile 
& Markdown. It’s coming along very nicely: I think I’m almost done! However I’ve 
run into a pattern with LPeg that I can’t seem to get right. Quoting from my 
command-prompt:

> = lpeg.match(lpeg.P'*' * lpeg.C((1 - lpeg.S'\r\n\f*')^1) * '*', '*emphasis*')
emphasis*

Did I mis-interpret the manual or is this a bug? Quoting part of the summary 
table from the LPeg manual:

	patt1 - patt2	Matches patt1 if patt2 does not match.

As you probably guessed I’m trying to match the asterisks and their content but 
only capture what’s between the asterisks. If I mis-interpreted the manual, do 
any of you know the correct way to capture what I have in mind? At the moment 
I’m using a post processing pass over the parse tree:

	token[2] = token[2]:match '^(.-)%*?$'

to trim trailing delimiters like the asterisk, double asterisk and several quote 
variants. I would however much prefer to match the correct input to begin with 
because it saves a lot of repetition and reduces the possibility of bugs.

Thanks in advance,

 - Peter Odding