htm2txt_adult168

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     htm2txt_adult168
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Tue Mar 11 05:03:49 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     convert html file download from adult168.com to text
   7 ###APPUSAGE:    [htmlfile...]
   8 ###APPEXAMPLE:  htm2txt_adult168 booksrc/1.shtml
   9 ###APPOPTION:
  10 use strict;
  11 use HTML::TreeBuilder;
  12 use lib $ENV{XR_PERL_MODULE_DIR};
  13 use MyPlace::Script::Usage qw/help_required/;
  14
  15 exit 0 if(help_required($0,@ARGV));
  16 use MyPlace::Filename qw/get_uniqname/;
  17 use MyPlace::HTML;
  18 use MyPlace::HTML::Convertor;
  19
  20
  21 my @flist;
  22
  23 if(@ARGV) {
  24     push @flist,@ARGV;
  25 }
  26 else {
  27     push @flist,"/dev/stdin";
  28 }
  29
  30 my $index=0;
  31 my $count=@flist;
  32 foreach my $fn (@flist) {
  33     my $title;
  34     my @src;
  35     my $dst;
  36
  37     $index++;
  38     next unless(-f $fn or $fn="/dev/stdin");
  39     open FI,"-|","cat '$fn' | gb2utf" or die("$!\n");
  40     while(<FI>) {
  41         push @src,$_;
  42         $title = get_title($_) unless($title)
  43     }
  44     close FI;
  45
  46     if(!$title) {
  47         $title = $fn;
  48         $title =~ s/^.*\///g;
  49         $title =~ s/\.(html|htm)$//;
  50     }
  51     else {
  52         $title =~ s/\s+.*$//;
  53         $title =~ s/\[.*$//g;
  54     }
  55     $dst = get_uniqname($title,".txt");
  56     print STDERR "[$index/$count]$fn -> $dst ...";
  57     if(! open FO,">",$dst) {
  58         print STDERR ("\n$!\n");
  59         next;
  60     }
  61     print FO $title,"\n","\n";
  62
  63     my @dst=@{text_from_string(\@src,undef,undef,["zoom"],undef)};
  64     my $blanks=0;
  65     foreach(@dst) {
  66         chomp;
  67         if(/^\s*$/) {
  68             $blanks++;
  69         }
  70         else {
  71             if($blanks) {
  72                 $_ .= "\n";
  73                 $blanks=0;
  74             }
  75             print FO $_;
  76         }
  77     }
  78     close FO;
  79     print STDERR "\n";
  80 }
  81 exit 0;
  82
  83