作者 gjp [perl] 2012-04-21 12:16 (点击下载)

#! /bin/env perl
# text.pl -- get HTML <title>..</title>
 
#TODO:
# * 支持在第二行处的标题:
# 	<title>
# 		This is title </title>
# * 支持去除标题开头的空格
#
#
#您能帮我改进一下吗？请将结果发到<gjp1120@gmail.com>
#十分感谢！
 
use strict;
use warnings;
 
mkdir "newfs";
 
my @files = glob "*.html";
 
foreach my $file (@files)
{
	do_file($file);
}
 
sub do_file
{
	my ($file_name) = shift @_;
	open FD, "<$file_name";
 
	foreach (<FD>){
		if (/<title>/i){
						chomp(my @t = split /<title>/i);
						$_ = $t[$#t];
						chomp(@t = split /</title>/i);
						$_ = $t[0];
						if(not defined $_){return;};
						chomp(@t = split /[-_]/i);
						$_ = $t[0];
						chomp(@t = split /[/i);
						$_ = $t[0];
 
						#print $_ . "n";
						change_name($file_name, $_);
		}
	}
	close FD;
}
 
sub change_name
{
	(my $fn, my $nfn) = @_;
 
	#print $nfn;
	rename $fn,"newfs/".$nfn.'.html';
}