If you use CGI.pm, then there's no need to get messy with regexps: the param and (self_)url function/methods are all you need ...

use CGI qw/ :standard /;

print header;

  # save requested slice
my $slice = param( 'slice' );

  # generate URLs to next 3 slices
for my $offset ( 1..3 ) {
    param( 'slice', $slice + $offset );
    print a( { -href => self_url }, query_string );
}

  # restore original (if needed)
param( 'slice', $slice );

__END__

Query:  slice=31&this=that
Output: slice=32&this=that
        slice=33&this=that
        slice=34&this=that

Query:  this=that&slice=49&that=this
Output: this=that&slice=50&that=this
        this=that&slice=51&that=this
        this=that&slice=52&that=this

Query:  this=that&that=this&slice=5
Output: this=that&that=this&slice=6
        this=that&that=this&slice=7
        this=that&that=this&slice=8
[download]

Season with input validation/error handling to taste. :)

    --k.

kanji is right, of course. CGI.pm will make your life much easier here. It deals with the problem you're facing, and with a hundred more that haven't come up yet and never will if you adopt it now...

To answer your question, for once, here's a modified version of what i use for paging back and forth. it normally lives in a List class, so this is a bit hacked out and has all its sanity-checks removed:

use CGI;
my $query = new CGI;

my $slice = $query->param('slice');
my $next_qs = build_qs($query, slice => $slice + 1);

sub build_qs {
    my ($query, %new_param) = @_;
    my %parameters = map { $_ => $query->param($_) } $query->param;
    $parameters{$_} = $new_param{$_} for keys %new_param;
    return join '&', 
           map { "$_=$parameters{$_}" } 
           keys %parameters;
}
[download]

In this example, $next_qs will contain the query string you need to append your script address to get the next page up.

build_qs($cgi_object, %hash) will return a query string which is exactly the same as the one supplied to the script except that any name => value pairs you supply in the %hash will be used instead of (or in addition to) the present input. the generalisation will be useful when you start wanting to change list order and length as well.

I'm sure there's a more elegant way to do the building part - someone will probably supply a one-liner that uses CGI.pm properly - but this works for me.

Solved it a different way in case anybody is curious. It handles all 3 scenarios:

$pre = substr ( $ENV{'REQUEST_URI'}, 0, index ( $ENV{'REQUEST_URI'}, "slice=" ) + 6 );
$post = substr ( $ENV{'REQUEST_URI'}, index ( $ENV{'REQUEST_URI'}, "slice=" ) + 6 );
if ( index ( $post, "&" ) >= 0 ) { 
    $post = substr ( $post, index ( $post, "&" ) ); 
} else { 
    $post = "";
}

I was expecting a convuluted regex but this seems to do the trick.

Thanx everyone.

======================
Sean Shrum
http://www.shrum.net

Does it have to be a regex. It seems straightforward, using "&" and then "=" to split the string.

untested code below:

@pairs = split( /\&/, $ENV{'REQUEST_URI'} );
[download]

Then go through the array, split the strings on '='.

Update:

CrazyInsomniac pointed out that my split was not esacping the '&'

<cite>Just a tongue-tied, twisted, earth-bound misfit. -- Pink Floyd</cite>

my %FORM = ();
my @QUERYARR = split(/\&/, $ENV{'REQUEST_URI'});
my $i;

for $i (0 .. $#QUERYARR) {
    my ($key,$val) = split(/=/, $QUERYARR[$i]);
    $val =~ tr/+/ /;
    $val =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C",hex($1))/eg;  #this t
+akes care of all the urlencoded data
    $FORM{$key} = $val;
}
[download]