RobotRules 和UserAgent来下载文件

#!/usr/bin/perl
​
use strict;
use warnings;
use WWW::RobotRules;
use LWP::UserAgent;
use HTTP::Request;
use HTTP::Response;
​
my $url = 'http://www.people.com.cn/';
my $agent = LWP::UserAgent->new;
​
# 创建一个RobotRules对象，用于检查网站的下载规则
my $robot_rules = WWW::RobotRules->new($url);
​
# 检查网站是否允许下载音频文件
if ($robot_rules->allowed('audio/*')) {
    print "网站允许下载音频文件。\n";
​
    # 使用LWP::UserAgent下载音频文件
    my $req = HTTP::Request->new(GET => $url);
    my $res = $agent->request($req);
​
    # 检查下载是否成功
    if ($res->is_success) {
        my $content = $res->content;
        my $filename = 'people_com_cn_audio.mp3'; # 默认保存文件名
​
        # 根据Content-Disposition头部信息获取实际文件名
        if ($res->header('Content-Disposition')) {
            my ($name) = $res->header('Content-Disposition') =~ /filename="([^"]+)/;
            $filename = $name if $name;
        }
​
        # 保存下载的音频文件
        open(my $output, '>', $filename) or die "Cannot open file: $!";
        print $output $content;
        close($output);
​
        print "音频文件下载完成，保存为 $filename。\n";
    } else {
        print "下载失败：" . $res->status_line . "\n";
    }
} else {
    print "网站不允许下载音频文件。\n";
}