#!/usr/local/bin/ruby
# amazon.rb
# ISBN-search -> detail-parse
# original; isbn2bib.rb for bk1    by takahashi
#           amazon.rb   for amazon by not
# ver.0.4  2002.3.21-2003.1.20
# ver.0.5  2003.5.20-2003.5.22
# ver.0.6  2003.11.10 adult対応
# ver.0.6a 2003.11.28 出版社情報の取り方修正
# ver.0.7  2003.12.22-2004.01.07 新cacheに変更
# ver.0.7a 2004.03.31 初動時imageを拾えてないのを修正
# ver.0.7b 2004.04.06 「4〜6週間」を拾えてないのを修正

#一部のお薦め（この商品に興味がある人は……でないのもの）等のデコードがまだ。
#descに文中リンクがある場合に未対応。今は除去してしまっている。

if __FILE__ == $0
  require 'cookie'
  require 'webagent'
  require 'cache'
  require 'cachebib'
end

class CacheBib_amazon < CacheBib
  begin
    require 'iconv'
  rescue LoadError
    begin
      require 'uconv'
    rescue LoadError
    end
  end
  attr_accessor :series_num, :limage, :edition, :recommend, :ranking
  
  def initialize ()
    super
    @shop_id = 'amz'
    @url_site = 'http://www.amazon.co.jp/'
    @affiliate_code = 'ipposjunkbox-22'
    @reviews = Array.new()
    @flag_review = true
    @adult_check = true
  end
  
  #===== ===== parts ===== =====
  def word_delivery ()
    delivery = ''
    flag = false
    if !@title.nil? # title-check -> no-info / get-info check
      flag = true
      if @delivery.nil?
        delivery = "ない"
      else
        if @delivery =~ /24/
          delivery = "24時間"
        elsif @delivery =~ /2/
          delivery = "2日"
        elsif @delivery =~ /3/
          delivery = "2〜3日"
        elsif @delivery =~ /取/
          delivery = "取寄せ"
        elsif @delivery =~ /1〜2週間/
          delivery = "1〜2週間"
        elsif @delivery =~ /4〜6週間/
          delivery = "4〜6週間"
        else
          delivery = "？"
        end
      end
    else
      message("WAR; maybe data-error, no-title.")
    end
    return delivery, flag
  end
  
  #===== ===== url ===== =====
  def url_isbnsearch ( isbn = @isbn )
    url_bibinfo(isbn)
  end
  
  def url_bibinfo ( id = @id )
    id = @isbn if id.nil?
    id = id.delete('-').upcase
    url = @url_site + "exec/obidos/ASIN/#{id}"
    url << "/ref_#{affiliate_code}" unless @affiliate_code.nil?
    url << '?val=authorized'        unless @adult_check.nil?
    return url
  end
  
  #キーワード検索主体。そのうち別途詳細検索用を用意した方がいいか？
  def url_search ( srch = 1, kywd = nil,  codes = {}, aid = @affiliate_code )
    if !kywd.nil?
      begin
        kywd = kywd.toeuc
        kywd = Iconv::iconv("UTF-8", "EUC-JP", kywd).to_s
      rescue NameError
        begin
          kywd = CGI::escape(kywd.toutf8)
        rescue
        end
      end
    end
    url = @url_site + "exec/obidos/external-search/?"
    url << "tag=#{affiliate_code}&" unless @affiliate_code.nil?
    url << "keyword=#{kywd}"
    return url
  end
  
  #===== ===== parse ===== =====
  def parse_bibinfo ( str )
    @id = @isbn.delete('-').upcase if @id.nil?
    #----- parse ----- #???keyできりわけてからparseする方法にそのうち変更???
#    if %r|本のデータ: (.+?)</title>|s =~ str
    if %r|Amazon.co.jp：(.+?)</title>|sm =~ str
      @title = $1.strip.gsub(/^本:/s,'').strip
    end
    
    if %r|シリーズ番号: (.+?)\n|s =~ str
      @series_num = $1
    end
    
    if %r|(<a href="/exec/obidos/search-handle-url/index=books-jp&field-author=[^\"]+">.+?<br>)| =~ str
      authors = $1.split(/<a href=/)
      authors.each { |line|
        if %r|"/exec/obidos/search-handle-url/index=books-jp&field-author=[^\"]+">(.+?)</a>| =~ line
          name = $1
          if line =~ /\(著\)/
            if @author.nil? then @author = '' else @author << "\t" end
            @author << name
          elsif line =~ /\(原著\)/
            if @e_author.nil? then @e_author = '' else @e_author << "\t" end
            @e_author << name
          elsif line =~ /\(翻訳\)/
            if @translator.nil? then @translator = '' else @translator << "\t" end
            @translator << name
          end
        end
      }
    end
    
    if %r|<b class="sans">.+?&nbsp;(.+?)</b>| =~ str
      @series = $1.gsub(/&nbsp;/,'').strip
    end #???label???
    
    if %r|<img src="http://images-jp.amazon.com/images/P/#{@id}.09.MZZZZZZZ.jpg" width=(\d+) height=(\d+)| =~ str
      @image = "http://images-jp.amazon.com/images/P/#{@id}.09.MZZZZZZZ.jpg"
      @image_width  = $1.to_i
      @image_height = $2.to_i
    end
    
    if %r|<a href="http://images-jp.amazon.com/images/P/#{@id}.09.LZZZZZZZ.jpg"| =~ str
      @limage= "http://images-jp.amazon.com/images/P/#{@id}.09.LZZZZZZZ.jpg"
    end
    
    if %r|価格：.*?<b class="price">(￥[\d,]+)</b>|mis =~ str
      @price = $1.strip
    end
    
    if %r|<b>発送可能時期：</b>([^<]*?)<br>|mis =~ str
      @delivery = $1.strip
    end
    
    if %r|<b>エディション：</b>([^<]*?)<br>|mis =~ str
      @edition = $1.strip
    end
    
    if %r|この商品に興味がある人は、こんな商品にも興味をもっています.+?<ul>(.+?)</ul>|mis =~ str
      list = $1
      recommend = []
      list.scan(%r|<li>.+?</li>|mi){ |line|
        if %r|/exec/obidos/ASIN/([\dX]+)/.+?<i>(.+?)</i></a>(.+?)</li>|mi =~ line
          recommend << [$1.strip, $2.strip, $3.strip]
        end
      }
      @recommend = recommend.flatten.join("\t") if recommend.size > 0
    end
    
#    if %r|書籍データ.+?<b>([^<]+)</b>\n(?: - (\d+) p)?|m =~ str
    if %r|書籍データ.+?<b>([^<]+)</b>\n(?:\s*?(\d+)\s*?p)?|mis =~ str
      edition = $1.strip.chomp(':')
      @page = " #{$2}p" unless $2.nil?
      if @edition.nil?
        @edition = edition
      elsif @edition != edition
        message("WAR; edition-missmatch: #{@edition} - #{edition}")
      end
    end
    
#    if %r|サイズ(.*?)\n|m =~ str #ブックカバー（文庫サイズ）に誤ヒット
    if %r|書籍データ.+?サイズ(.*?)\n|mis =~ str
      pubsize = $1.gsub(/\s+/, ' ').gsub(/<[^>]+>/, '').strip
      @pubsize = pubsize if pubsize != '(cm):'
    end
    
#    if %r|^([^>]+)\n ; \nISBN|m =~ str then
    if %r|<b>出版社: </b>(.*?)\n|m =~ str
      @publisher = $1.strip
    end
    
    if %r|ISBN:\s*([\dX]+)| =~ str
      @isbn = $1.strip
    end
    
    if %r|^\(([\d\/]+)\)| =~ str
      @pubdate = $1
    end
    
    if %r|<b>Amazon.co.jp 売上ランキング：</b>([^<]+)|mis =~ str
      @ranking = $1.strip
    end
    
#    if %r|<i>内容（「.+?」データベースより）</i></b> <br>  </font> +([^<]+)<BR>|s =~ str
    if %r|<span class="serif">(.+?)</span>|mi =~ str
      @desc = $1.gsub(/&nbsp;/i, '').gsub(/<br>/i, "\t").gsub(/<p>/i, "\t").gsub(/<[^>]+>/i, '').strip
    end #時々文中リンクがあるが、今は除く
    
    if %r|<b class="h1">目次</b>(.+?)<hr|mi =~ str
      @index = $1.gsub(/\s+/, ' ').gsub(/<br>/i, "\t").gsub(/<[^>]+>/, '').strip
    end
    
    if str =~ %r|<a name="cust-reviews"><b class="h1">カスタマーレビュー</b><br></a>(.*)|mis
      data = $1
    end
#    reviewsplit = data.to_s.split("投稿者")
    reviewsplit = data.to_s.split(/stars-.+?\.gif/)
    
    reviewlist = []
    reviewsplit.each{ |column|
      if column =~ /=review-id value=(\d+)/
        custid = $1
        reviewlist << custid
        if column =~ %r|レビュアー：.*?<b>(.*?)</b>|mis
          custid = $1.gsub(/<[^>]+>/, '').strip.gsub(/\(プロフィールを見る\)$/s, '').strip
          custid = "(noname)" if custid.size == 0
          reviewlist << custid
        end
      end
    }
    if reviewlist.size > 0
      @reviewlist = reviewlist.join("\t")
      @review_num = reviewlist.size / 2 #ID, name, ID, name...
    end
    
    if @flag_review
      reviewsplit.each{ |column|
        review = {}
        if column =~ /.+?alt=([^>]+?)>/
          key = $1.strip
          review["value"] = key if key !~ /^"/
        end
        if column =~ /=review-id value=(\d+)/
          review["review_id"] = $1.strip
        end
        if column =~ /^.*?<b>([^>]*?)<\/b>, ([\d\/]*)/
          review["title"] = $1.strip
          review["date"] = $2
        end
        if column =~ %r|レビュアー：.*?<b>(.*?)</b>(.+?)</table>(.+?)<form|mis
          reviewer, place, content = $1, $2, $3
          reviewer = reviewer.gsub(/<[^>]+>/, '').strip.gsub(/\(プロフィールを見る\)$/s, '').strip
          reviewer = "(noname)" if reviewer.size == 0
          review["reviewer"] = reviewer
          review["place"] = place.gsub(/<[^>]+>/, '').gsub(/　/, ' ').strip
          content.gsub!(/<br>/mi,"\n")
          content.gsub!(/\n\s+/,"\n")
          content.gsub!(/\s+/," ")
          content.gsub!(/\n+/,"\n")
          content.gsub!(/<p>/mi,"\n")
          content.strip!
          content.gsub!(/\n/,"\t")
          review["content"] = content #???もうちょっと修正が要る???
        end
        @reviews << review if review.size > 0
      }
    end
  end
  
  #===== ===== get/set info ===== =====
  def preset_info #===== set local-value from @info (from cache) =====
    super
    @id = @isbn.delete('-').upcase #@info["id"]
  end
  
  def set_info #===== set book-info(@info) from local-value =====
    super
    #----- set info add ----- #amazon独自の情報類
    @info["series_num"] = @series_num
    @info["limage"]     = @limage
    @info["edition"]    = @edition
    @info["recommend"]  = @recommend
    @info["ranking"]    = @ranking
  end
end

if __FILE__ == $0
  p CacheBib_amazon.new.get_data('4-15-010739-4')
end