Re: dfa, awk

[http://dummy.us.eu.org/robert (robert)]

 > From: Leonid Leibman <http://www.raqia.com/~leonidl>
 > Date: Tue, 04 Sep 2001 09:12:06 -0400
 >
 > ---Executing: html-to-ascii
 > Hi, Robert -- first of all (if you're still interested in the subject) I
 > learnt that the algorithm for conversion of
 > a regular expression directly to DFA (without the intermediate NFA) is
 > much more efficient than the
 > classical Thompson algorithm I (we) used. Somehow the need of e-closure
 > for empty transitions is
 > eliminated from the picture by some trick. All this is in the next
 > section of the Dragon book. Seems like
 > the commercial products use this algorithm as well.

That's good to know.

 > Second -- a question -- how can one use backreferences in awk (in sed
 > one does s/a\(b*\)a\1/.....).
 > Can I do something of this sort in awk? For example can I accomplish
 > something like
 > 
 > /a\(b*\)a\1/ { print \1}
 > 
 > or maybe something simpler like
 > 
 > /a\(b*\)a/ {print \1}
 > 
 > More generally, is there some mechanism like this in awk?-- 

I don't know of any way.  Can't sed be used?  If I had to do this, I'd
probably just use Perl.

 > Leonid Leibman