#!/usr/bin/env ruby
#
# Author: IKARASHI Yoshinori <yoosee at gmail dot com >
# Created: 2005-09-17 21:45:01 JST
# Modified: 2007-10-01 23:4426:
# Licence:  GPLv2
# $Id$
#
=begin
= mixi2gmail description

mixi2gmail feeds mixi diary to some e-mail account, such as
gmail.

= configuration
this program requires configuration file at $HOME/.mixi/mixirc .
It should be including parameters as follows:

username = foo@example.com      # mixi account e-mail address
password = xxxxxxxx             # mixi login password
mailaddress = foo@gmail.com     # e-mail address mail to send
mailfrom  = mixi@init.org       # dummy email From address
smtpserver = localhost          # it dosen't work yet because of 
                                # mechanize override problem
excludelist = 1111, 34567       # skip list
watchlist = 98765, 43210, 2222  # watch list who are not your friend

= website

http://trac.yoosee.net/mixi2gmail/

=end

$LOAD_PATH.unshift("/home/yoshi/work/web/mixi")

require 'nkf'
require 'mechanize'
require 'net/smtp'
require 'time'
require 'optparse'

module Mixi

  class Config
    attr_reader :config
    def initialize(rcfile)
      @rcfile = rcfile
      @config = Hash.new
      config_file_parser
    end

    def config_file_parser(rcfile=@rcfile)
      return nil unless File.exist? rcfile
      File.open(rcfile) do |file|
        file.each do |line|
          key, value = line.chomp.split(/\s*=\s*/)
          if /,/ =~ value
            values = value.split(/\s*,\s*/)
            value = values
          end
          @config[key] = value
        end
      end
    end

    def command_line_parser
      opt = OptionParser.new
    end
  end

  IMAGEBLOCK_REGEXP = '<td[^>]*?><a href="javascript:void\(\d\);" onClick="MM_openBrWindow\(\'(show_diary_picture.pl\?owner_id=\d+&id=\d+&number=\d+)\',\'pict\',\'width=\d+,height=\d+,toolbar=no,scrollbars=yes,left=\d,top=\d\'\);"><img src="(http://ic\d+.mixi.jp/p/[0-9a-f]+/[0-9a-f]+/diary/\d+/\d+/[0-9_]+s\.jpg)" alt="" /></a></td>'
  IMAGEBLOCK_CONTAINER_REGEXP = '<div class="diaryPhoto">
<table cellpadding="0" cellspacing="0" border="0" summary="写真"><tr>
\s*
</tr></table>
</div>' 

  USERNAME_REGEXP = '<div class="diaryTitleFriend clearfix">\s*<h2>(.+)さんの日記</h2>\s*</div>'
  DIARY_REGEXP = '<!--\[viewDiaryBox\]-->
<div class="viewDiaryBox">
\s*
<div class="listDiaryTitle">
<dl class="clearfix">
<dt>(.+)
<img src=".+?" alt=".+?" width="\d+" height="\d+" /></dt>
<dd>(\d{4}年\d{2}月\d{2}日)(\d{2}:\d{2})</dd>
</dl>
</div>
\s*
<div class="txtconfirmArea">
\s*
<div id="diary_body">(.+?)</div>'

  COMMENT_REGEXP = '<span class="commentTitleName"><a href="(show_friend.pl\?id=\d+)">(.+?)</a>
</span>
\s*
<span class="commentTitleDate">(\d{4}年\d{2}月\d{2}日)&nbsp;(\d{2}:\d{2})</span>
\s*
</dt>
\s*
<dd>
(.+?)
</dd>
</dl>
</div>'

  class Diary
    attr_accessor :author, :datetime, :title, :imagepages, :images,
      :body, :url, :comments
    def initialize
      @comments = Array.new
      @imagepages = Array.new
    end

    def mkdatetime date, time
      if(/(\d{4})\D+(\d{2})\D+(\d{2})/) =~ date
        year, month, date = $1, $2, $3
      end
      if(/(\d{2}):(\d{2})/) =~ time
        hour, minute = $1, $2
      end

      Time.local(year,month,date,hour,minute)
    end

    def parse html
      html = NKF::nkf('-ed', html)
      author, datetime, title, body = nil, nil, nil, nil
      @author = nil
      html.gsub!(/#{IMAGEBLOCK_REGEXP}/) {
        imagepage_uri = $1
        next unless imagepage_uri
        @imagepages.push imagepage_uri
        puts "  [i] image #{imagepage_uri}"
        ""
      }
			html.gsub!(/#{IMAGEBLOCK_CONTAINER_REGEXP}/m,'')
      if /#{USERNAME_REGEXP}/m =~ html
        author = $1
      end

      if /#{DIARY_REGEXP}/m =~ html
        title, date, time, body = $1, $2, $3, $4
        puts "  + #{author}: #{title} (#{date} #{time})"
      else
        puts "  + failed to parse diary of #{author}"
        return false
      end
      @author = author
      @title  = title
      @body   = body
      @datetime = mkdatetime(date, time)
      
      html.gsub(/#{COMMENT_REGEXP}/m){
        url, author, date, time, body = $1, $2, $3, $4, $5
        @comments.push(DiaryComment.new(author, mkdatetime(date,time), url, body))
      }
      return true
    end

    def html
      output = '<html><body>'
      output << "<h1><a href=\"http://mixi.jp/#{@url}\">#{@title}</a></h1> by #{@author} at #{@datetime.strftime('%Y-%m-%d %H:%M')}<br /><br />"
#      @images.each do |image|
#        output << "<a href='#{image}'><img src=\"#{image}\" alt=\"[IMAGE]\" /></a> "
#      end
      output << "<p>#{@body}</p><hr />"
      @comments.each do |comment|
        output << "<h3>#{comment.author} at #{comment.datetime.strftime('%Y-%m-%d %H:%M')}</h3>#{comment.body}<hr />"
      end
      output << '</body></html>'
      output
    end
  end

  class DiaryComment
    attr_accessor :author, :datetime, :url, :body
    def initialize(author, datetime, url, body)
      @author, @datetime, @url, @body = author, datetime, url, body
    end
  end

  class DiaryFetcher
    attr_accessor :username, :password, :lastidfile, :logfile, :debug,
                  :excludelist, :wait
    attr_reader :lastid
    def initialize(username, password)
      @username, @password = username, password
      @logfile = ENV['HOME'] + '/.mixi/' + 'mixi-access.log'
      @lastidfile = ENV['HOME'] + '/.mixi/' + 'lastid.log'
      @excludelist = Array.new
      @wait  = 3
      login
      @lastid = File.open(@lastidfile).read.to_i
      @maxid  = @lastid
      @debug = false
    end

    def login
      @agent = WWW::Mechanize.new {|a| a.log = Logger.new(@logfile) }
			@agent.user_agent_alias = 'Windows Mozilla'
			puts "access to http://mixi.jp as #{@agent.user_agent}"
      page = @agent.get('http://mixi.jp/')
#      page = @agent.get('mixi.html')
      form = page.forms[0]
#      form.fields.find {|f| p f}
      form.fields.find {|f| f.name == 'email'}.value = @username
      form.fields.find {|f| f.name == 'password'}.value = @password
      form.fields.find {|f| f.name == 'next_url'}.value = '/home.pl'
      page = @agent.submit(form, form.buttons.first)
      puts "login to mixi as #{@username}." if @debug
      if /url=([^"])"/ =~ page.body
        link = 'http://mixi.jp' + $1.to_s
        @agent.get(link)
      end
      sleep @wait if @wait > 0
    end

    def fetch_user_diary ownerid
      listurl = "http://mixi.jp/list_diary.pl?id=#{ownerid}"
      page = @agent.get(listurl)
      diary = Array.new
      page.links.each do |link|
        if /続きを読む/ =~ link.node.text &&
            /view_diary.pl\?id=(\d+)&owner_id=(\d+)/ =~ link.href
          diaryid = $1.to_i
          @maxid = diaryid if @maxid < diaryid
          if diaryid <= @lastid
#            puts "pre-fetched diary: #{link.href}" if @debug
            next
          end
          sleep @wait if @wait > 0
          mixidiary = Diary.new
          mixidiary.url = link.href
          mixidiary.parse(@agent.get(link.href).body)
          diary.push mixidiary
        end
      end
      return diary
    end

    def fetch_friend_diary
      listurl = 'http://mixi.jp/new_friend_diary.pl'
      page = @agent.get(listurl)

      list = Array.new
      page.body.gsub(/view_diary.pl\?id=(\d+)&owner_id=(\d+)/) do |href|
        diaryid, ownerid = $1.to_i, $2.to_i
        @maxid = diaryid if @maxid < diaryid
        if @lastid.to_i >= diaryid
#          puts "pre-fetched diary: #{href}" if @debug
          next
        end
        if @excludelist.include? ownerid.to_s # to_s is temporaly...
          puts "skip #{$1} because of exclude list" if @debug
          next
        end
        list.push ownerid
      end

      diary = Array.new
      list.reverse.each do |ownerid|
        sleep @wait if @wait > 0
        puts "fetching diary: id=#{ownerid}" if @debug
        userdiary = fetch_user_diary(ownerid)
				diary << userdiary unless userdiary.empty?
      end
      return diary.flatten
    end

    def fetch_image imagepage_uri
      return if imagepage_uri.empty?
#      puts "       + try to get #{imagepage_uri}"
      image = nil
      page = @agent.get("http://mixi.jp/" + imagepage_uri)
      page.body.gsub(/<img src="(http:\/\/[^"]+\.jpg)"[^>]*>/i) {
        src = $1
        puts "    -> fetch image: #{src}"
        image = @agent.get(src).body
      }
      return image
    end

    def close
      File.open(@lastidfile, 'w') do |file|
        file.write(@maxid)
      end
    end
  end
end

def send_mail(text, subject, datetime, mail_from, recipients, images)
  header = ''
  imagebody = ''
  unless images.empty?
    boundary = '__mixi2gmail_123456789ABCDEFGHIJ__'
    header << "Content-Type: multipart/mixed; boundary=\"#{boundary}\""
    content = ''
    tmp = text
    text = "--#{boundary}\n"
    text << "Content-Type: text/html;\n\n"
    text << tmp
    text << "\n\n"
    images.each do |image|
      filename = Time.now.to_i.to_s + '_' + rand(1000).to_i.to_s + '.jpg'
      content = "--#{boundary}\n"
#      content << "Content-Type: application/octed-stream;\n"
      content << "Content-Type: image/jpeg;\n"
      content << " name=\"#{filename}\"\n"
      content << "Content-Transfer-Encoding: base64\n"
      content << "Content-Disposition: attachment;\n"
      content << " filename=\"#{filename}\"\n\n"
      content << [image].pack('m')
      content << "\n\n"
      imagebody << content
    end
    imagebody << "--#{boundary}\n"
  end
  text << imagebody
  tempfile = Tempfile.open('/tmp')
  tempfile.write NKF.nkf('-j',text)
  tempfile.close
  system("cat #{tempfile.path} | mailx -s \"#{NKF.nkf('-M', subject)}\" -a 'X-ML-Name: mixi' -a 'From: mixi diary <#{mail_from}>' -a 'Content-Type: text/html; charset=\"iso-2021-jp\"' -a 'Date: #{datetime.rfc2822}' -a '#{header}' #{recipients}")
  tempfile.unlink
  return
# following (net/smtp) dosen't work because of mechanize's override
=begin
   smtpserver = 'localhost'
   message = "From: #{mail_from}
To: #{recipients}
Subject: #{NKF::nkf('-M', subject)}
Date: #{Time.now.rfc2822}
MIME-Version: 1.0
X-ML-Name: mixi
Content-Type: text/plain; charset=\"iso-2022-jp\" 
Content-Transfer-Encoding: 7bit

#{NKF.nkf('-j',text)}
"
   Net::SMTP.start(smtpserver, 25) do |smtp|
     smtp.send_message(message, mail_from, recipients)
   end
=end
end

rcfile = "#{ENV['HOME']}/.mixi/mixirc"
config = Mixi::Config.new(rcfile)
unless config
	puts "configuration file #{rcfile} not found."
	exit
end

agent = Mixi::DiaryFetcher.new(config.config['username'], 
                               config.config['password'])
agent.debug = true
agent.excludelist = config.config['excludelist']
agent.fetch_friend_diary.reverse.each do |diary|
	begin
    images = Array.new
    diary.imagepages.each do |imagepage|
      images.push agent.fetch_image(imagepage)
      sleep 2
    end
    send_mail(diary.html, diary.author+'さんの日記', diary.datetime,
            config.config['mailfrom'], config.config['mailaddress'],
             images)
	rescue
		puts "Error: #{diary.url if diary.url} => #{$!}:#{$@}"
		p diary
	end
end

config.config['watchlist'].each do |ownerid|
  agent.fetch_user_diary(ownerid).each do |diary|
	  begin
      images = Array.new
      diary.imagepages.each do |imagepage|
        images.push agent.fetch_image(imagepage)
        sleep 2
      end
      send_mail(diary.html, diary.author+'さんの日記', diary.datetime,
              config.config['mailfrom'], config.config['mailaddress'],
               images)
	  rescue
	  	puts "Error: #{diary.url if diary.url} => #{$!}:#{$@}"
	  end
  end
end
agent.close