Path: utzoo!attcan!uunet!salt.acc.com!ucsd!ogicse!hakanson
From: hakanson@ogicse.ogi.edu (Marion Hakanson)
Newsgroups: comp.lang.perl
Subject: Re: User Definable Character Classes
Message-ID: <8084@ogicse.ogi.edu>
Date: 17 Mar 90 18:58:05 GMT
References: <15253@bfmny0.UU.NET> <7420@jpl-devvax.JPL.NASA.GOV> <260176EA.C58@tct.uucp>
Organization: Oregon Graduate Institute (formerly OGC), Beaverton, OR
Lines: 52

In article <260176EA.C58@tct.uucp> chip@tct.uucp (Chip Salzenberg) writes:
>What we could use here is a full-blown lexical analyis engine
>integrated into the Perl language.
>
>For example, the RCS file format is an ASCII stream with "@"
>delimeters, where "@@" means a literal "@".  I've often wondered how
>to write Perl to interpret such a file without calling getc thousands
>of times.

Take it from someone who's been down this road.  You DON'T want to call
getc lots of times.  I converted my perl "dnslex" to C, and the C version
ran probably 300 times faster.  But putting the lex-er in a separate
program works quite well, esp. with Perl's nice way of opening pipes.

However, the @/@@ problem isn't so tough, as long as you don't have to
worry about backslash-escapes (where the backslash could be escaped).
Even that can be done in absence of other quoting mechanisms, but it
is not pretty (see below).

Here's a routine I wrote to split on a comma (instead of an @), with
it doubled for a literal comma.  Maybe it will give an idea....

# The arg may be of the form 'part1,part2', where ',' is
# the first un-doubled comma (later commas are not processed).

sub commasplit {
    local ($_) = @_;
    local ($first,$secnd);

    $first = '';
    $secnd = '';
    
    commasplit: while ( /,/ ) {
        $first .= $`;	# before the comma
        $_ = $';	# and after it

        if ( s/^,// ) {	# turn double into a single & continue
            $first .= ',';
        } else {	# make the split
            $secnd = $_;
            $_ = '';	# remainder goes above
            last commasplit;
        }
    }
    $first .= $_;	# in case no single comma was found
    ($first,$secnd);
}


-- 
Marion Hakanson         Domain: hakanson@cse.ogi.edu
                        UUCP  : {hp-pcd,tektronix}!ogicse!hakanson