sinabook_h2t

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     sinabook_h2t
   3 ###APPAUTHOR:   duel
   4 ###APPDATE:     Wed Mar 12 01:44:35 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     convert html file download from book.sina.com.cn to plain-text
   7 ###APPUSAGE:    [html file]
   8 ###APPEXAMPLE:  sinabook_h2t foo.htm\n\tcat foo.htm | sinabook_h2t
   9 use strict;
  10 $ARGV[0]="-h" unless(@ARGV);
  11 foreach(@ARGV){
  12     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  13 }
  14
  15 my $processor="htm2txt";
  16 my $rootid = '-s "<font id=\"zoom\" class=f14><p>" -e "<table align=right><tr><td>"';
  17 #my $rootid = '-r contTxt,artibody,article -s "<div id=article>" -e "返回目录"';
  18 if(@ARGV) {
  19     open FI,"-|","iconv -f gb2312 -t utf8 '$ARGV[0]' | $processor $rootid";
  20 }
  21 else {
  22     open FI,"-|","$processor $rootid";
  23 }
  24 while(<FI>) {
  25     s/^.*(新浪|相关链接|读书频道).*$//g;
  26     print;
  27 }
  28 close FI;