#! /usr/local/bin/perl
# =====================================================================
#  http://www.searchterms.com/~searchterms/index.html
#  を再帰的に吸い上げたあと、そのトップディレクトリでこれを実行すれば
#  たぶん searchterms の上位ランキングを普通のリスト形式にして出力。
# =====================================================================
#system("wget -m --no-parent http://www.searchterms.com/~searchterms/index.html");

if(open(O,"| sort -n")) { select O; }
@files = grep(/\.html$/, split(/[\r\n]+/,`find . -type f`));
foreach $file ( @files ) {
  if($file =~ /\/\_/) { next; }
  if(open(I,"$file")) {
    while($line = <I>) {
      @lis = split(/[\r\n]+/, $line);
      foreach ( @lis ) {
        if(/\<P ALIGN\=\"CENTER\"\>\<B\>[^\<]+\<\/B\>\<\/TD\>/) {
          s/\<[^\<]+\>//g;
          if(/^\#/) { $num = $'; next; }
          print "$num\t$_\n";
        }
      }
    }
    close(I);
  }
}
exit;