#!/usr/bin/perl

package MT::Plugins::TextHatena;

use vars qw($VERSION);
$VERSION = 0.01;

use strict;
use Jcode;
use Text::Hatena;

use MT;
use MT::Blog;
use MT::Entry;
use MT::Template::Context;

MT->add_text_filter(text_hatena => {
  label => 'MT::Text::Hatena',
  on_format => sub {
    my ($text, $ctx) = @_;
    if($text eq ''){
      return "";
    }
    my $entry = $ctx->stash('entry');
    my $permalink = "";
    if ($entry && $entry->id) {
      $permalink = $entry->permalink;
    }
    my $parser = Text::Hatena->new(
      permalink => $permalink,
      );
    my $html = $parser->parse($text);
    parseLinks($parser->html);
  },
});

sub parseLinks{
  my $source = shift;
  my @lines  = split(/\n/, $source);
  foreach my $line(@lines){
    # さまざまなパターンについて分岐
    $line =~ s!(
      # src="URL", href="url"
      ((src|href|cite)=[\'\"](http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*[\'\"])
      |
      # []URL[]
      (\[\](http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*\[\])
      |
      # [URL]
      (\[(http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*(:image)?\])
      |
      # [URL]
      (\[(http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*(:title=.*)?\])
      |
      # [URL]
      ((http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*)
      |
      # [map:WORDs]
      #map:x139.6983y35.6514
      #[map:渋谷区鉢山町]
      #[map:150-0035]
      #[map:株式会社はてな]
      (\[(map):[^\]]+\])
      |
      # [search:WORDs]
      (\[(search):((keyword|question|asin):)?[^\]]+\])
      |
      # [google:WORDs]
      (\[(google):((image|news):)?[^\]]+\])
      |
      # [amazon:WORDs]
      (\[(amazon):[^\]]+\])
      |
      # [rakuten:WORDs]
      (\[(rakuten):[^\]]+\])
      |
      # id:hogehoge:yyyymmdd
      (id:\w+(:\d+)?)
      |
      # g:hugahuga:id:hogehoge:yyyymmdd
      (g:\w+(:id:\w+(:\d+)?)?)
      |
      # ?:id:hogehoge:〜
      #a:id:sample
      #b:id:sample,b:id:sample:20050707,b:id:sample:favorite,b:id:sample:asin
      #d:id:hatenadiary,d:id:hatenadiary:20040510
      #i:id:sample
      #r:id:sample
      #f:id:hatenadiary:20041007101545j:image,f:id:hatenadiary:20041007101545j:image:small
      #f:id:hatenadiary:20041007101545j:image:w50,f:id:hatenadiary:favorite
      ((a|b|i|r|f|map):id:\w+(:\w+)?)
      |
      # [keyword:WORDs]
      #[keyword:はてな]
      #[keyword:はてな:graph]
      #[keyword:はてな:graph:accessrank:3d]
      (\[(keyword):[^\]]+\])
      |
      #[g:hatena:keyword:はてなダイアリー利用可能タグ]
      (\[g:\w+:(keyword):[^\]]+\])
      |
      )!&toHtml($1)!igex;
    #idea:182,idea:182:title
    #graph:id:sample
    #[graph:id:sample:しなもんの体重]
    #[graph:id:sample:しなもんの体重:image]
    #jan:4981254610640
    #ean:4981254610640
    #jan:4981254610640:title
    #[jan:4981254610640:title=この商品]
    #jan:4981254610640:image
    #jan:4981254610640:barcode
    #id:hatenadiary:about
    #id:hatenadiary:archive
    #id:hatenadiary:archive:200508
    #question:1096559040
    #question:1096559040:title
    #question:1092057258:detail
    #uestion:1092057258:image
    #question:1092057258:image:small
  }
  join("\n",@lines);
}

sub toHtml{
  my $line = shift;
  # src="URL", href="url"
  if($line =~ /(src|href|cite)=[\'\"](http|https|ftp|mailto):([^\x00-\x20()<>\x7F-\xFF])*[\'\"]/){
    return $line;
  }
  # []URL[]
  if($line=~/\[\]((http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*)\[\]/){
    return $1;
  }
  # [URL:image]
  if($line=~/\[((http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*)(:image)?\]/){
    return qq(<img src="$1" alt="" title="$1" />);
  }
  # [URL]
  if($line=~/\[((http|https|ftp|mailto|isbn|asin|ISBN|ASIN):([^\x00-\x20()<>\x7F-\xFF])*)(:title=(.*))?\]/){
    return ($4 eq '') ? qq(<a href="$1">$1</a>) : qq(<a href="$1" title="$5">$5</a>);
  }
  # [URL]
  if($line=~/((http|https|ftp|mailto):([^\x00-\x20()<>\x7F-\xFF])*)/){
    return qq(<a href="$1" title="$1">$1</a>);
  }
  # [map:WORDs]
  #未対応：map:x139.6983y35.6514
  #[map:渋谷区鉢山町],[map:150-0035],[map:株式会社はてな]
  if($line=~/\[(map:([^\]]+))\]/){
    my $encoded = toUtf8($2);
    return qq(<a href="http://map.hatena.ne.jp/?word=${encoded}">$1</a>);
  }
  # [search:WORDs]
  if($line=~/\[(search:((keyword|question|asin):)?([^\]]+))\]/){
    my $u = 'http://search.hatena.ne.jp';
    if($3 ne ''){
      $u .= ($3 eq 'keyword')  ? '/keyword'     :
            ($3 eq 'question') ? '/questsearch' :
            ($3 eq 'asin')     ? '/asinsearch'  : '';
    }
    else{
      $u .= '/keyword';
    }
    $u .= '?word=' . toEuc($4);
    return qq(<a href="$u">$1</a>);
  }
  # [google:WORDs]
  if($line=~/\[(google:((image|news):)?([^\]]+))\]/){
    my $u='http://www.google.com/search';
    my $encoded = toEuc($4);
    if($2){
      $u = ($3 eq 'image') ? 'http://images.google.com/images' :
           ($3 eq 'news')  ? 'http://news.google.com/news'     : '';
    }
    return qq(<a href="$u?q=${encoded}&ie=euc-jp&oe=euc-jp">$1</a>);
  }
  # [ASIN:xxx]
  if($line=~/((isbn|asin|ISBN|ASIN):(\w+)(:(detail|image)(:(small|large))?)?)/){
    return qq(<a href="http://d.hatena.ne.jp/asin/$3" title="$1">$1</a>);
  }
  # [amazon:WORDs]
  if($line=~/\[(amazon:([^\]]+))\]/){
    my $encoded = toEuc($2);
    return qq(<a href="http://www.amazon.co.jp/exec/obidos/external-search?mode=blended&tag=hsjjp-22&encoding-string-jp=${encoded}&keyword=${encoded}">$1</a>);
  }
  # [rakuten:WORDs]
  if($line=~/\[(rakuten:([^\]]+))\]/){
    my $encoded = toEuc($2);
    return qq(<a href="http://pt.afl.rakuten.co.jp/c/002e8f0a.89099887/?sv=2&v=3&p=0&sitem=${encoded}">$1</a>);
  }
  # [keyword:WORDs]
  #(\[(keyword):[^\]]+\])
  if($line=~/\[(keyword:([^\]]+))\]/){
    my $encoded = toEuc($2);
    return qq(<a href="http://d.hatena.ne.jp/keyword/${encoded}">$1</a>);
  }
  # id:hogehoge:yyyymmdd
  #d:id:hatenadiary,d:id:hatenadiary:20040510
  if($line =~ /((d:)?id:(\w+)(:(\d+))?)/){
    return qq(<a href="http://d.hatena.ne.jp/$3/$6">$1</a>);
  }
  # g:hugahuga:id:hogehoge:yyyymmdd
  if($line =~ /(g:(\w+)(:id:(\w+)(:(\d+))?))/){
    return qq(<a href="http://$2.g.hatena.ne.jp/$4/$6">$1</a>);
  }
  #b:id:sample,b:id:sample:20050707,b:id:sample:favorite,b:id:sample:asin
  if($line =~ /\[?(b:id:(\w+)(:(\w+))?)\]?/){
    return qq(<a href="http://b.hatena.ne.jp/$2/$4">$1</a>);
  }
  #b:keyword:xxx
  if($line=~/\[(b:keyword:([^\]]+))\]/){
    my $encoded = toEuc($2);
    return qq(<a href="http://b.hatena.ne.jp/keyword/${encoded}">$1</a>);
  }
  #idea:999
  if($line =~ /(idea:(\d+))/){
    return qq(<a href="http://i.hatena.ne.jp/idea/$2">$1</a>);
  }
  #?:id:hogehoge:〜
  if($line =~ /((a|b|d|f|i|r|f|map):id:(\w+))/){
    return qq(<a href="http://$2.hatena.ne.jp/$3">$1</a>);
  }
  #f:id:hatenadiary:20041007101545j:image,f:id:hatenadiary:20041007101545j:image:small
  #f:id:hatenadiary:20041007101545j:image:w50,f:id:hatenadiary:favorite
  #[g:hatena:keyword:はてなダイアリー利用可能タグ]
  #(\[g:\w+:(keyword):[^\]]+\])
  $line;
}

sub toEuc(){
  my $src = shift;
  my $enc = Jcode->new($src)->euc;
  $enc =~ s/(\W)/'%' . unpack('H2', $1)/eg;
  $enc =~ tr/ /+/;
  $enc;
}
sub toUtf8(){
  my $src = shift;
  my $enc = Jcode->new($src)->utf8;
  $enc =~ s/(\W)/'%' . unpack('H2', $1)/eg;
  $enc =~ tr/ /+/;
  $enc;
}
1;