htmllinks

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     htmllinks
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Thu Mar 27 22:43:15 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     strip links from html source
   7 ###APPUSAGE:    [BaseURL]
   8 ###APPEXAMPLE:  cat index.htm | htmllinks www.google.com
   9 ###APPOPTION:
  10 use strict;
  11 use URI;
  12 use HTML::TreeBuilder;
  13 foreach(@ARGV){
  14     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  15 }
  16
  17 my $base=shift;
  18 if($base) {
  19     $base=URI->new($base);
  20     $base="http://$base" unless($base->scheme);
  21 }
  22
  23 my $tree=HTML::TreeBuilder->new();
  24 while(<STDIN>) {
  25     $tree->parse($_);
  26 }
  27 $tree->eof();
  28 if($base) {
  29     foreach my $node($tree->find("a","link")){
  30         my $url=$node->attr("href");
  31         print URI->new_abs($url,$base),"\n" if($url);
  32     }
  33     foreach my $node($tree->find("img")){
  34         my $url=$node->attr("src");
  35         print URI->new_abs($url,$base),"\n" if($url);
  36     }
  37 }
  38 else {
  39     foreach my $node($tree->find("a","link")){
  40         my $url=$node->attr("href");
  41         print ("$url\n") if($url);
  42     }
  43     foreach my $node($tree->find("img")){
  44         my $url=$node->attr("src");
  45         print ("$url\n") if($url);
  46     }
  47 }
  48 $tree->delete();
  49 exit 0