WGet_VIKSOE

   1 #!/usr/bin/perl -w
   2
   3 #http://www.viksoe.dk/code/windowless1.htm
   4
   5
   6 sub file_name {
   7     my $r = shift;
   8     $r =~ s/.*\/\///;
   9     $r =~ s/^[^\/]*\///;
  10     $r =~ s/\?/@/g;
  11     return $r;
  12
  13 }
  14
  15 sub article_name {
  16     my $r = shift;
  17     $r =~ s/.*[\\\/]//g;
  18     return $r;
  19 }
  20
  21 my @exclude = (
  22     "/script/Jobs",
  23     "/script/Catalog",
  24     "/script/Membership",
  25     "/Forums",
  26     "/script/Forums",
  27     "/info",
  28     "/webservices",
  29     "/script/Ratings",
  30     "/script/Catalog",
  31 );
  32 sub start_wget {
  33     my $url = shift;
  34     my @wget = ("wget","-N","-nH","-r","-l","1","-A","htm,css,js,zip,rar,exe,gif,jpg,jpeg","-k","-E","-t",3);
  35     if(@exclude) {
  36         push @wget,"-X";
  37         push @wget,join(",",@exclude);
  38     }
  39     return (system(@wget,$url)==0);
  40 }
  41
  42 sub get_article {
  43     my $url=shift;
  44     return unless($url);
  45     $url =~ s/\.aspx.*/\.aspx\?display=PrintAll/;
  46     my $file_name = file_name($url);
  47     my $article_name = article_name($file_name);
  48     if(start_wget($url)) {
  49         open FO,">",$article_name;
  50         print FO <<EOF;
  51     <frameset>
  52         <frame border=0 src="$file_name">
  53     </frameset>
  54 EOF
  55     print STDERR "$article_name ==> $file_name\n";
  56         return 1,$url,"OK";
  57     }
  58
  59     return undef,$url,"Download failed";
  60 }
  61
  62 if(@ARGV) {
  63     foreach(@ARGV) {
  64         my @r = get_article($_);
  65         if($r[0]) {
  66             print STDERR "$r[2]:$r[1]\n";
  67         }
  68         else {
  69             print STDERR "$r[2]:$r[1]\n";
  70         }
  71     }
  72 }
  73 else {
  74     print STDERR $0,"\n";
  75     print STDERR "Usage:\n\t$0 (Viksoe URLs)\n";
  76 }
  77