递归wget文件的一个perl程序(不完备，暂没时间改)

功能：将目录开始，下载匹配所有的*.gz, *.bz2, *.zip文件；
缺陷：处理递归目录可能会有问题
TODO：改进，将$site改为参数；文件后缀改为参数；

#!/usr/bin/perl
# (C)icymoon
my $site = "";
my $curr_site;
my @wget_res = ();

sub get_url($)
{
        my $sitei = $_[0];
        my @wget_res = `wget $_[0] 2>&1`;
        print("wget $_[0]\n");
        my $curr_index;
        if($wget_res[$#wget_res-1] =~ /(index.html.*)\'\s+saved/) {
                $curr_index = $1;
                my @site_info = `cat ./$curr_index`;
                foreach my $line (@site_info) {
                        if($line =~ /a\s+href\=\"(.*\/)\"\>/) {
                                get_url("$sitei\/$1");
                        }
                        elsif ($line =~ /a\s+href\=\"(.*\.gz)\"\>/)
                        {
                                `wget $sitei\/$1 2>&1`;
                                if($? != 0) {
                                        print FD "wget $sitei\/$1 failed!\n";
                                }
                                print("wget $sitei\/$1    $?\n");
                        }
                        elsif ($line =~ /a\s+href\=\"(.*\.bz2)\"\>/)
                        {
                                `wget $sitei\/$1 2>&1`;
                                if($? != 0) {
                                        print FD "wget $sitei\/$1 failed!\n";
                                }
                                print("wget $sitei\/$1    $?\n");
                        }
                        elsif ($line =~ /a\s+href\=\"(.*\.zip)\"\>/)
                        {
                                `wget $sitei\/$1 2>&1`;
                                if($? != 0) {
                                        print FD "wget $sitei\/$1 failed!\n";
                                }
                                print("wget $sitei\/$1    $?\n");
                        }
                }
        }
}
if(!open(FD,"> wget_log"))
{
        printf("Error in open log file\n");
        exit("1");
}
get_url($site);
close(FD);