I have a somewhat complex (for me, anyway) routine here that SHOULD go out and grab a random picture from a remote site. Here is the code.

  %seen = (); @uniq = grep { ! $seen{$_} ++ } @z;
        my $query = q|SELECT weburl,datecode,idnum FROM maindb WHERE i
+dnum IN (| . join(',', map { $dbh->quote($_) } @uniq) . q|)|;
        my($sth) = $dbh->prepare($query);
        $sth->execute || die("Could not execute!");
        my $array_ref = $sth->fetchall_arrayref();
        
  foreach my $row (@$array_ref) {
  my ( $url, $dc, $idn ) = @$row;

  if ($url){

  $ua = new LWP::UserAgent;
  $ua->agent ('Mozilla/4.0 (compatible; MSIE 5.03; Windows 95)');

  my @imgs = ();
  sub callback {
     my($tag, %attr) = @_;
     return if $tag ne 'img';
     push(@imgs, values %attr);
  }

  $p = HTML::LinkExtor->new(\&callback);
  $res = $ua->request(HTTP::Request->new(GET => $url), sub {$p->parse(
+$_[0])});
  my $base = $res->base;
  @imgs = map { $_ = url($_, $base)->abs; } @imgs;
  @foo = grep /\.jpg/i, @imgs; 
  $randompic = $foo[rand @foo];

 $req1 = new HTTP::Request 'GET' => $randompic; 
 my $img_response = $ua->request($req1);
 $content = $img_response->content; 


        open(TGPMAIN,">$cat_html/featured/$idn\.jpg") || print "Can't 
+open gall REASON: ($!)\n";
        print TGPMAIN "$content";
        close(TGPMAIN);  
        
$sql = "INSERT INTO photofour VALUES('$idn','$dc','$randompic')";
$dbh->do($sql);
}
}
[download]

For the first picture, it works great. Everything works as expected at first, but if @z contains more than one value only the first is actually processed correctly. For instance, @z might have 123 and 456 in it. The script will select the URL from the DB with 123 as its ID. The script will then spider that URL and retrieve a list of .jpg images available there. It will randomly choose one and download it to a local directroy naming it 123.jpg. After that is over with, the "photos" table will have inserted into it the ID number, a date code, and original URL of the pic. The trouble would come when trying to spider the next URL (the one with ID 456 in this case). The script creates a blank image with the proper name and into the DB is entered the ID number and date code, the original URL is not present. Can anyone spot why this will only work the first time through? I have been on this for several hours and it's starting to confuse me more and more as time goes by.