use File::Find;
use Data::Dumper;

die "usage: html2db.pl dirname" unless ( $ARGV[0] );

## MT の import ファイルを import.txt に書き出し。MT の import 機能使って取り込んでね
open my $oh, '>', 'import.txt' or die $!;
my %datahash;
my $firstflg = 0;
my @tags;
find( { wanted => \&filter, no_chdir => 1 }, $ARGV[0] );
close $oh;

## tag sql データ生成 を別ファイルに書き出し。コレを元に手動でタグを入力してね
open $oh, '>', 'tags.txt' or die $!;
print $oh Dumper( [@tags] );
close $oh;

## -----------------------------------------------------------------------------
## 拡張子が html のファイルを検索してパースします
## -----------------------------------------------------------------------------
sub filter {
    my $name;
    my $dir;
    eval {
        $name = $File::Find::name;
        $dir  = $File::Find::dir;
        return if ( $name !~ /^(.+\.html)$/ );
        my $finename = $1;
        &parser($name);
    };
    if ($@) {
        warn "$name : $@ \n";
    }
}

## -----------------------------------------------------------------------------
## html パーサー。各自のテンプレにあわせて正規表現変えてくんろ
## -----------------------------------------------------------------------------
sub parser() {
    my $filename = shift;
    warn $filename;

    my $MTEntryTitle;
    my $MTCategory;
    my $MTEntryBody;
    my $MTEntryMore;	## 残念ながら MTEntryBody に吸収されちゃう。
    my $MTEntryDate;
    my $MTEntryAuthor;
    my @trackbacks;
    my @comments;

    open my $fh, '<', $filename or die $!;
    my $html = do{ local $/; <$fh> };

	## title 復旧。RDF から抽出してみる
    if ( $html =~ m!dc:title="(.+?)"!ms ) {
        $MTEntryTitle = $1;
    }
	## body 復旧。
    if ( $html =~ m!<div\sclass="entry-body">\n\s*(.+?)\n\s*</div>[\r\n\s]+?<div\sclass="entry-body2">!ms ) {
        $MTEntryBody = $1;
    }
	## category 復旧。
    if ( $html =~ m!class="entry-navi-header">Categories:.+?<a\shref.+?">(.+?)</a>!ms ) {
        $MTCategory = $1;
	## entrydate 復旧。
    if ( $html =~ m!class="entry-footer">.+?class="post-footers">\s*([^\|]+?)\s*\|.+?(\d\d)/(\d\d)/(\d\d).+?(\d+):(\d+)</span>!ms ) {
        my $time = $5;
        my $time2 = sprintf( "%02d", $6 );
        if ( $time > 12 ) {
            $time = sprintf( "%02d", $time - 12 );
            $MTEntryDate = qq{$3/$4/20$2 $time:$time2:00 PM};
        }
        else {
            $time = sprintf( "%02d", $time );
            $MTEntryDate = qq{$3/$4/20$2 $time:$time2:00 AM};
        }
        ## 日付が解析できなければ、2007/01/01 00:00:00 で初期化。手抜き。
        $MTEntryAuthor = $1;
    }
	## tag 復旧。便宜上100個までしか復旧しない
    for ( 0 .. 100 ) {
        if ( $html =~ m!rel="tag">(.+?)</a>!msg ) {
            push @tags, { title => $MTEntryTitle, tag => $1 };
        }
        else {
            last;
        }
    }
    }
	## comment 復旧。便宜上100個までしか復旧しない。名前、書き込み日時、内容のみ復旧。後は手抜き
    for ( 0 .. 100 ) {
        if ( $html =~ m!class="comment"\sid="comment-(\d+)">.+?class="comment-content">\n\s*(.+?)\n\s*</div>[\r\n\s]+?<p\sclass="comment-footer">\s*([^\|]+?)\s*\|.+?(\d\d)/(\d+)/(\d+).+?(\d\d):(\d\d)!msg ) {
            my $date  = qq{$5/$6/20$4 $7:$8:00};
            my $time  = $7;
            my $time2 = sprintf( "%02d", $8 );
            if ( $time > 12 ) {
                $time = sprintf( "%02d", $time - 12 );
                $date = qq{$5/$6/20$4 $time:$time2:00 PM};
            }
            else {
                $time = sprintf( "%02d", $time );
                $date = qq{$5/$6/20$4 $time:$time2:00 AM};
            }
            push @comments, { id => $1, body => $2, auther => $3, date => $date };
        }
        else {
            last;
        }
    }
	## trackbacks 復旧。便宜上100個までしか復旧しない。タイトル、URL,書き込み日時のみ復旧。後は手抜き
    for ( 0 .. 100 ) {
        if ( $html =~ m!class="trackback-list-item"><a\shref="(.+?)"\srel="nofollow\snofollow">(.+?)</a></li>!msg ) {
            my $date = qq{10/07/2007 07:00:00 PM};
            push @trackbacks, { url => $1, title => $2, date => $date };
        }
        else {
            last;
        }
    }
    ## コメントのデータ生成。もうホント場当たりなスクリプト。
    my $commnum = scalar @comments;
    my $pingnum = scalar @trackbacks;
    my $commstr = '';
    my $pingstr = '';
    for (@comments) {
        $commstr .= qq{-----
COMMENT:
AUTHOR: $_->{auther}
EMAIL: 
IP: 127.0.0.1
URL: 
DATE: $_->{date}
$_->{body}
};
    }
    if ( $commstr eq '' ) {
        $commstr = qq{-----
COMMENT:
};
    }
    ## trackbacksのデータ生成。もうホント場当たりなスクリプト。
    for (@trackbacks) {
        $pingstr .= qq{-----
PING:
TITLE: $_->{title}
URL: $_->{url}
IP: 127.0.0.1
BLOG NAME: ... orz
DATE: $_->{date}
oh! no!! ... orz
};
    }
    if ( $pingstr eq '' ) {
        $pingstr = qq{-----
PING:
};
    }
    ## export データ生成前に、もし同じ日時、同じタイトルなら無視することにしてみる
    $MTEntryDate = qq{01/01/2007 0:00:00 AM} if ( $MTEntryDate eq '' );
    $MTEntryBody = qq{... orz} if ( $MTEntryBody eq '' );
    return if ( $datahash{"$MTEntryTitle:$MTEntryDate"} );
    $datahash{"$MTEntryTitle:$MTEntryDate"} = 1;
	## export データ生成。
	print $oh qq{\n--------\n} if ( $firstflg );
	$firstflg = 1;
    print $oh qq{AUTHOR: $MTEntryAuthor
TITLE: $MTEntryTitle
STATUS: Publish
ALLOW COMMENTS: 1
CONVERT BREAKS: __default__
ALLOW PINGS: 1
PRIMARY CATEGORY: $MTCategory
CATEGORY: $MTCategory

DATE: $MTEntryDate
-----
BODY:
$MTEntryBody
-----
EXTENDED BODY:

-----
EXCERPT:

-----
KEYWORDS:

$commstr
$pingstr
-----

};

}

1;

__END__