#!/usr/bin/perl

use strict;
use warnings;

use Getopt::Long;
use Pod::Usage;
use Path::Class qw(dir file);
use meon::Web::Config;
use XML::Chain qw(xc);
use DateTime;
use URI;
use URI::Escape qw(uri_escape);
use List::Util qw(max);
use 5.010;

my $google_sitemap_image_ns = 'http://www.google.com/schemas/sitemap-image/1.1';

exit main();

sub main {
    my $help;
    my $dst_domain;
    my $notify_email;
    my $author;
    my $changefreq = 'monthly';
    GetOptions(
        'help|h'       => \$help,
        'hostname=s'   => \$dst_domain,
        'changefreq=s' => \$changefreq,
    ) or pod2usage;
    pod2usage if $help;
    pod2usage unless defined $dst_domain;

    my $hostname_folder = meon::Web::Config->hostname_to_folder($dst_domain);
    my $dst_hostname_dir =
        dir(meon::Web::SPc->srvdir, 'www', 'meon-web', $hostname_folder, 'content',);
    die 'no such hostname ' . $dst_domain unless $hostname_folder;

    my $sitemap = xc(sitemap_xml())->set_io_any($dst_hostname_dir->file('sitemap.xml'));
    $dst_hostname_dir->recurse(
        callback => sub {
            my ($file) = @_;
            return if $file->is_dir;
            return if $file !~ m/\.xml$/;
            my $file_mtime = $file->stat->mtime;
            my $rel_file = $file->stringify;
            $rel_file =~ s/(index)?\.xml$//;
            $rel_file =~ s/^$dst_hostname_dir//;
            return if $rel_file eq '/sitemap';

            my $content_xml = xc($file);
            $content_xml->reg_global_ns('w' => 'http://web.meon.eu/');
            return unless $content_xml->find('/w:page/w:content')->count;

            my $url =
                URI->new('http://'
                    . $dst_domain
                    . join('/', map {uri_escape($_)} file($rel_file)->components));
            my $url_el = $sitemap->c('url');
            $url_el->c('loc')->t($url);
            $url_el->c('lastmod')
                ->t(DateTime->from_epoch(epoch => $file_mtime)->strftime('%Y-%m-%d'));
            $url_el->c('changefreq')->t($changefreq);

            # priority based on distance from today
            my $mtime_days = int((time - $file_mtime) / (24 * 60 * 60) + 0.5) || 1;
            $url_el->c('priority')->t(sprintf('%.2f', max(0.01, 1 / $mtime_days)));

            if (my $thumb = $content_xml->find('//w:img-thumb')->text_content) {
                my @thumb_comp;
                if ($thumb =~ m{^/}) {
                    @thumb_comp = file($thumb)->components;
                }
                else {
                    @thumb_comp = (file($rel_file)->components, $thumb);
                }
                my $thumb_uri =
                    URI->new('http://' . $dst_domain . join('/', map {uri_escape($_)} @thumb_comp));
                $url_el->c('image', xmlns => $google_sitemap_image_ns)->c('loc')->t($thumb_uri);
            }
            $sitemap->t("\n");
        }
    );
    $sitemap->store;

    return 0;
}

sub sitemap_xml {
    return XML::LibXML->load_xml(string => <<'__XML_TEMPLATE__');
<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
</urlset>
__XML_TEMPLATE__
}

=head1 NAME

meon-web-generate-sitemap - generate content sitemap.xml

=head1 SYNOPSIS

    meon-web-generate-sitemap --hostname domain
        --hostname domain   where to store timeline entries
        --changefreq $      one of always hourly daily weekly monthly yearly never

=head1 DESCRIPTION


=cut