html2unix

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     html2unix
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Wed Apr 28 02:26:38 2010
   5 ###APPVER:      0.1
   6 ###APPDESC:     html2unix convert \"href\" \"src\" to lowercase
   7 ###APPUSAGE:
   8 ###APPEXAMPLE:  html2unix
   9 ###APPOPTION:
  10 use strict;
  11
  12 #ENV variable MUST be defined somewhere,
  13 #FOR perl to search modules from,
  14 #OR nothing will work
  15 use lib $ENV{XR_PERL_MODULE_DIR};
  16
  17 use MyPlace::Script::Usage qw/help_required help_even_empty/;
  18 #exit 0 if(help_required($0,@ARGV));
  19 exit 0 if(help_even_empty($0,@ARGV));
  20 use HTML::TreeBuilder;
  21
  22
  23 sub lc_links {
  24     my $tree = shift;
  25     foreach my $attr (qw/src href/) {
  26         my @elms = $tree->look_down($attr,qr/./);
  27         map {$_->attr($attr,lc($_->attr($attr)))} @elms;
  28     }
  29     return $tree;
  30 }
  31
  32 foreach my $file (@ARGV) {
  33     if($file eq '-') {
  34         my $tree = HTML::TreeBuilder->new_from_content(<STDIN>);
  35         lc_links($tree);
  36         print FO $tree->as_HTML();
  37         $tree->delete();
  38     }
  39     elsif(-f $file and $file =~ m/\.(?:html|htm|xml|xhtml)$/) {
  40         print STDERR "Converting $file...";
  41         my $tree = HTML::TreeBuilder->new_from_file($file);
  42         lc_links($tree);
  43         open FO,">".$file or die("$!\n");
  44         print FO $tree->as_HTML();
  45         close FO;
  46         $tree->delete();
  47         print STDERR "\t [OK]\n";
  48     }
  49     else {
  50         print STDERR "\"$file\" Ingored.\n";
  51     }
  52 }
  53
  54
  55
  56
  57
  58