grapDLSite

#! /usr/bin/perl

use 5.014;
use autodie;
#use LWP::Simple;
use Web::Query;
use JSON;
use utf8;
use Encode;
use Data::Dumper;

use DJVoiceConfig;

binmode(STDIN, ':encoding(utf8)');
binmode(STDOUT, ':encoding(utf8)');
binmode(STDERR, ':encoding(utf8)');

#my $ARGV[0] = 'RJ128196';

if(! ($ARGV[0] =~ /(RJ[0-9]{6})/) ){
    die 'need RJ###### in $ARGV[0]';
}
my $target = $1;
my $working_dir = $DJVoiceConfig::WORKING_DIR;
my $file = "$working_dir/$target.html";

my $dlsite = 'http://www.dlsite.com/maniax/work/=/product_id/';

system("curl -s -o $file $dlsite$target.html") if( ! -e $file);
my $html = Web::Query->new_from_file($file);
#->find('div.head dt')
#    ->find('.work_story')
my $title = $html->find('div.base_title_br')->find('a')->text();
my $discount = $html->find('span.icon_lead_01')->text();
Encode::_utf8_on($title);
Encode::_utf8_on($discount);
$title =~ s/\W*$discount//;

my %genre;
$html->find('table')->filter(sub { return $_->attr('id') == 'work_outline'; })
    ->find('span')->each(sub {
        my $text = $_->text();
        Encode::_utf8_on($text);
        $genre{ $_->attr('class') } = $text;
    });

#print Dumper(@genre);
my $main_genre;
my $flag_adult = 0;
while ( my($key, $value) = each %genre ){
    $main_genre = '同人音声' if($key eq 'icon_SOU');
    $main_genre = 'CG+ノベル' if($key eq 'icon_IN2');
    $main_genre = '動画' if($key eq 'icon_MOV');
    $flag_adult = 1 if ($key eq 'icon_ADL');
}

my %tag;
$html->find('div.main_genre')->find('a')
    ->each(sub {
            $_ = $_->text();
            Encode::_utf8_on($_);
            $tag{$_}++;
            #my $i = shift;
            #printf("%d %s\n", $i+1, $_->text());
            #printf("%d %s\n", $i+1, $_->as_html());
    });
$html->find('table.reviewer_most_genre')->find('span')
    ->each(sub {
            $_ = $_->text();
            s/...\d+...//;
            Encode::_utf8_on($_);
            $tag{$_}++;
    });
$tag{'全年齢'}++ if(!$flag_adult);


my $image = $html->find('meta')
    ->filter(sub {
            return $_->attr('name') =~ /twitter:image:src/;
    })->attr('content');

my %circle;
$circle{'id'} = $html->find('span')
    ->filter(sub {
            return $_->attr('class') =~ /maker_name/;
    })->find('a')->attr('href');
$circle{'id'} =~ s/^.*maker_id\///;
$circle{'id'} =~ s/\.html//;

$circle{'name'} = $html->find('span')
    ->filter(sub {
            return $_->attr('class') =~ /maker_name/;
    })->find('a')->text();
Encode::_utf8_on( $circle{'name'} );

my $date = $html->find('table')
    ->filter(sub {
            return $_->attr('id') =~ /work_outline/;
    })->find('a')->filter(sub {
            return $_->attr('href') =~ /year/;
    })->attr('href');
$date =~ s/^.*\/year\///;
$date =~ s/\/cyear.*$//;
$date =~ s/\/mon\/|\/day\//-/g;

#my $i = 1;
my %vocal;
my $vocal_str;
my $has_vocal_tag = 0;
$html->find('table')->filter(sub { return $_->attr('id') == 'work_outline'; })
    ->find('tr')->filter(sub {
        $_ = $_->find('th')->text();
        Encode::_utf8_on($_);
        if(/声優/){
            $has_vocal_tag = 1;
            return 1;
        }
        return 0;
    })->find('td')->each(sub{
        $_ = $_->text();
        Encode::_utf8_on($_);
#        s/\s//;
#        s/\// /;
        $vocal_str = $_;
    });

#my $content = $html->find('div.work_article')
if(!$has_vocal_tag){
    $html->find('div.work_article')
        ->filter(sub {
                return 1 if $_->has_class('work_story');
                return 0;
        })->each(sub {
                $_ = $_->text();
                Encode::_utf8_on($_);
                if(/(ボイス)|(VOICE)|(voice)|(CV)|(声優)|(キャスト)/){
                    s/^.*$+\W+//s;
                    s/\s.*$//s;
                    s/様$|http.*$//;
                    s/\W+$//;
                    $vocal{$_}++;
                }
        });
    $vocal_str = join ('/', keys(%vocal));
}

my %output = (
    'id'            =>    $target,
    'title'         =>    $title,
    'circle_id'     =>    $circle{'id'},
    'circle_name'   =>    $circle{'name'},
    'image'         =>    $image,
    'date'          =>    $date,
    'tag'           =>    join (' ', keys(%tag)),
    'vocal'         =>    $vocal_str,
    'genre'         =>    $main_genre,
);
#say $content;
my $outputUtf8 = encode_json \%output;
Encode::_utf8_on($outputUtf8);
print $outputUtf8;