[原创代码] [Perl][LWP-Simple]批量获取微软官方主题包下载地址

=info
    Code by: ViANG/523066680
    E-Mail : 523066680@163.com
    Date   : 2014-01
=cut

use Encode;
use LWP::Simple;
use utf8;
my $main="http://windows.microsoft.com";
my $themepage="http://windows.microsoft.com/zh-cn/windows/themes";

binmode(STDOUT,':encoding(gbk)');
binmode(STDERR,':encoding(gbk)');

my @all;
my $title;
@all=getpage($themepage,"themes.htm");
$spec="最新主题";
foreach (@all) {
    if ($_=~/${spec}/) {
        $title=$_;
    }
}
if (not defined $title) {
    print "\n无法找到和主题类目有关的信息。请检查信息来源\n";
    exit();
}

my @items=split('a href',$title);
shift @items;
my (@c_name,@link);
my $i=0;
foreach (@items) {
    $_=~/="(.*)" class.*title="(.*)" data/;
    $c_name[$i]=$2;
    $link[$i++]=$main . $1;
}

# 获取标题直到 "全景（双显示器）" 为止
my $endmark="全景（双显示器）";
my $sum;
open ALLURL,">","linkofall.txt";
FO: foreach (0..($i-1)) {
    @all=getpage($link[$_], $c_name[$_].".txt");
    print ALLURL encode("utf-8","# $c_name[$_]:\n");
    foreach my $code (@all) {
        if ($code=~/download.*\..*themepack/) {
            $sum=analyse($code);
        }
    }
    print ALLURL encode("utf-8","# 共计 $sum 个链接\n\n");
    last FO if ($c_name[$_] eq $endmark);
}
close ALLURL;
print "\nOVER";
<STDIN>;
exit();

sub getpage {
    my ($link, $name)=($_[0],$_[1]);
    $name=encode('gb2312',$name);
    if ( not -e $name ) {
        print "file not exist, downloading";
        getstore($link,$name) or die "$!";
    }
    open READ,"<:utf8",$name or die "can not open file, $!";
    my @all=<READ>;
    close READ;
    return @all;
}

sub analyse {
    my @all=split(">下载<",$_[0]);
    my $sum=0;
    foreach (@all) {
        if (/"(http:\/\/download.*\..*themepack)"$/) {
            $sum++;
            print ALLURL $1,"\n";
        }
    }
    return $sum;
}
复制代码