Adapting parenthesis in regexps

cyzza has asked for the wisdom of the Perl Monks concerning the following question:

Hey all, I am currently writing a postfix log parser, a reasonable example is in $string.
a whole heap of key=value pairs, easy enough.
but I want to see if I can optimise what I already have done:

my $string = 'to=<adam.clark@ngv.vic.gov.au>, relay=monet1.ngv.vic.gov
+.au[10.10.10.20]:25, delay=0.54, delays=0.06/0.02/0/0.46, dsn=2.0.0, 
+status=sent ';

$LogLineHash = qr {
        ^
        ([^=]*)=<?(.*?)>?,?\s+)
        (.*?)
        $
}xi;

while ( $string ){
        print "String: $string\n";
        ( $junk , $key , $value, $string  )
                = split( /$LogLineHash/ , $string );

        print "Key: $key\nValue: $value\nLeft Over: $string\n\n";
        $Hash{$key}=$value;
}

while ( my ($key, $value) = each(%Hash) ) {
        print "$key => $value\n";
    }
[download]

Which gets me:

String: to=<adam.clark@ngv.vic.gov.au>, relay=monet1.ngv.vic.gov.au[10
+.10.10.20]:25, delay=0.54, delays=0.06/0.02/0/0.46, dsn=2.0.0, status
+=sent
Key: to
Value: adam.clark@ngv.vic.gov.au
Left Over: relay=monet1.ngv.vic.gov.au[10.10.10.20]:25, delay=0.54, de
+lays=0.06/0.02/0/0.46, dsn=2.0.0, status=sent

String: relay=monet1.ngv.vic.gov.au[10.10.10.20]:25, delay=0.54, delay
+s=0.06/0.02/0/0.46, dsn=2.0.0, status=sent
Key: relay
Value: monet1.ngv.vic.gov.au[10.10.10.20]:25
Left Over: delay=0.54, delays=0.06/0.02/0/0.46, dsn=2.0.0, status=sent

String: delay=0.54, delays=0.06/0.02/0/0.46, dsn=2.0.0, status=sent
Key: delay
Value: 0.54
Left Over: delays=0.06/0.02/0/0.46, dsn=2.0.0, status=sent

String: delays=0.06/0.02/0/0.46, dsn=2.0.0, status=sent
Key: delays
Value: 0.06/0.02/0/0.46
Left Over: dsn=2.0.0, status=sent

String: dsn=2.0.0, status=sent
Key: dsn
Value: 2.0.0
Left Over: status=sent

String: status=sent
Key: status
Value: sent
Left Over:

relay => monet1.ngv.vic.gov.au[10.10.10.20]:25
to => adam.clark@ngv.vic.gov.au
dsn => 2.0.0
status => sent
delay => 0.54
delays => 0.06/0.02/0/0.46
[download]

I was hoping that I could get my regex to dynamically grab all the key value pairs at once with a (?: )+ style grouping such that my regex would be:

^(?:([^=]*)=<?(.*?)>?,?\s+)+$
[download]

essentially grabbing arbitrary number of key value pairs, but it's not to be as:

Sample code:

my @bits = split( /^(?:([^=]*)=<?(.*?)>?,?\s+)+$/ , $string );
foreach ( @bits ){
    print "$_\n";
}
[download]

gives me:

status
sent
[download]

which is the last key value pair.
Is what I want to do possible?

also, can someone fill me in on when using split(), the first array element is always nothing. hence my $junk variable.

Comment on Adapting parenthesis in regexps Select or Download Code

Replies are listed 'Best First'.
Re: Adapting parenthesis in regexps by ikegami (Patriarch) on Mar 29, 2007 at 12:59 UTC
`split` works best when you have a list joined by a seperator, and you split on the seperator. `sub unbracket { for (my $s = @_ ? $_[0] : $_) { s/^<//; s/>$//; return $_; } } my %hash = map unbracket, map { split /=/ } split /, /, $_; # Or just # my %hash = map unbracket, split /, \|=/, $_;` [download]	[reply] [d/l] [select]
Re: Adapting parenthesis in regexps by Fletch (Bishop) on Mar 29, 2007 at 12:54 UTC
Why not just split on `/,\s/` then parse out `/([^=]+)=(.)/`? Update: And as to your question about the empty element, when your split regexp contains capturing parens the matched delimiters are returned as well as the delimited fields.	[reply] [d/l] [select]
Re: Adapting parenthesis in regexps by Ionitor (Scribe) on Mar 29, 2007 at 13:05 UTC
Or, you can just skip the use of split all together... `my %bits = $string =~ /([^=]+)=<?(.+?)>?\s(?:,\s\|$)/g;` Edit: added clause to grab last pair	[reply] [d/l]
Re^2: Adapting parenthesis in regexps by Ionitor (Scribe) on Mar 29, 2007 at 13:55 UTC
And, because I'm always looking to practice my regex-fu, here's a version with all greedy matches, which tend to be faster--I timed a roughly 40% speed increase. `my %bits = $string =~ / ([^=]+) # Key =<? # Seperator ([^,>]+ (?> >[^,>]+)) # Value >?(?:,\s\|\s*$) # Termination /gx;` [download]	[reply] [d/l]
Re: Adapting parenthesis in regexps by bobf (Monsignor) on Mar 29, 2007 at 14:48 UTC
a whole heap of key=value pairs, easy enough. but I want to see if I can optimise what I already have done I guess that depends on how you define "optimise". That could mean reusing existing code (thereby optimizing development time) unless there is a reason not to (such as computational speed, which is another thing that can be optimized). Since your input is a bunch of "key=value" pairs, you could use one of the many config modules that parse ini-like files. `use warnings; use strict; use Config::General; my $string = 'to=<adam.clark@ngv.vic.gov.au>, relay=monet1.ngv.vic.gov +.au[10.10.10.20]:25, delay=0.54, delays=0.06/0.02/0/0.46, dsn=2.0.0, +status=sent '; $string =~ s/, /\n/g; my $conf = Config::General->new( -String => $string ); my %hash = $conf->getall(); while( my ( $k, $v ) = each %hash ) { print "$k => $v\n"; }` [download] Prints: `to => <adam.clark@ngv.vic.gov.au> relay => monet1.ngv.vic.gov.au[10.10.10.20]:25 dsn => 2.0.0 status => sent delay => 0.54 delays => 0.06/0.02/0/0.46` [download]	[reply] [d/l] [select]