htm2txt

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     htm2txt
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Tue Mar 11 05:03:49 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     convert html file download from book.sina.bom.cn to txt
   7 ###APPUSAGE:    [htmlfile] [[-r rootId,[rootId,...]] [-f filter,[filter,...]] [-s regexp -e regexp]]
   8 ###APPEXAMPLE:  htm2txt booksrc/1.shtml "content,article" "sina.com" \n\tcat booksrc/1.shtml | htm2txt
   9 ###APPOPTION:   -r:rootIds|-f:filters|-s:starting mark|-e ending mark
  10 use strict;
  11 use HTML::TreeBuilder;
  12 use lib $ENV{XR_PERL_MODULE_DIR};
  13 use MyPlace::Script::Usage qw/help_required/;
  14 use MyPlace::HTML::Convertor;
  15
  16 exit 0 if(help_required($0,@ARGV));
  17
  18 my $fn;
  19 my @rootid = ();
  20 my @filter = ();
  21 my $start;
  22 my $end;
  23 my $what="";
  24
  25 if(@ARGV) {
  26     foreach my $opt (@ARGV) {
  27         if($opt =~ /^-[rfse]$/) {
  28             $what=$opt;
  29             next;
  30         }
  31         else {
  32             if($what eq "-r") {
  33                 push @rootid,split(",",$opt);
  34             }
  35             elsif($what eq "-f") {
  36                 push @filter,split(",",$opt);
  37             }
  38             elsif($what eq "-s") {
  39                 $start = $opt;
  40             }
  41             elsif($what eq "-e") {
  42                 $end = $opt;
  43             }
  44             else {
  45                 $fn=$opt;
  46             }
  47         }
  48     }
  49     if(!$fn) {
  50         $fn="/dev/stdin";
  51     }
  52 }
  53 else {
  54     $fn="/dev/stdin";
  55 }
  56
  57 open FI,"<",$fn or die("$!\n");
  58 my @src=<FI>;
  59 close FI;
  60
  61 my @result = @{text_from_string(\@src,$start,$end,\@rootid,\@filter)};
  62 print @result;
  63
  64