#!/usr/bin/env ruby
# yahoo.rb(books) Yahoo!JAPAN BOOKSはeS! Booksとの提携によります。
# ISBN-search -> detail-parse
# original; isbn2bib.rb for bk1 by takahashi
# ver.0.1 2003.1.19-2003.1.20
# ver.0.2 2004.01.05-2004.01.05 新cacheに変更

#読者レビュー（代表5評／別頁）
#「みんなの書店」レビュー（別頁）
#検索時のtableからの情報抽出
#以上、やり残し

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

class CacheBib_yahoo < CacheBib
  attr_accessor :user
  attr_accessor :review_point, :reviews_id
#  attr_accessor :ownerreviewlist
  
  def initialize ()
    super
    @shop_id = 'yho'
    @url_site = 'http://books.yahoo.co.jp/' 
    @reviews = Array.new()
    @flag_review = 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 =~ /1日〜3日/
          delivery = "1〜3日"
        elsif @delivery =~ /3日〜6日/ #??? 存在しない？
          delivery = "3〜6日"
        elsif @delivery =~ /3日〜3週間/
          delivery = "3〜21日"
        elsif @delivery =~ /お取り扱いできません/
          delivery = "不可"
        else
          delivery = "？"
        end
      end
    else
      message("WAR; maybe data-error, no-title.")
    end
    return delivery, flag
  end
  
  #===== ===== url ===== =====
  def url_isbnsearch ( isbn = @isbn )
    url = @url_site + "bin/search_key?p=#{isbn}"
    return url
  end
  
  def url_bibinfo ( id = @id )
    url = @url_site + "bin/detail?id=#{id}"
    return url
  end
  
  #===== ===== parse ===== =====
  def parse_isbnsearch ( str )
    count = []
    str.scan(%r|<a href="/bin/detail\?id=(\d+)">(.*?)</a>|){
      count << [$1, $2] unless $1.nil?
    }
    if count.uniq.size > 1
      message("WAR; ISBN-search hit multi-column. ISBN; #{@isbn}")
    end
    @id, @title = count.first
    return count
  end
  
  def parse_bibinfo ( str )
    #----- set key and data ------
    mode_tbl = {
      :image     => '<b>書籍画像</b>',
      :publisher => '<b>出版社名</b>',
      :title     => '<b>書籍名</b>',
      :label     => '<b>シリーズ名</b>',
      :author    => '<b>著者名</b>',
      :pubdate   => '<b>出版年月　</b>',
      :size      => '<b>ページ数・版型　</b>',
      :isbn      => '<b>ISBNコード</b>',
      :price     => '<b>価格</b>',
      :stock     => 'esbooksカート',
      :desc      => '<b>内容</b>',
      :index     => '<b>目次 </b>',
      :user      => '<b>読者対象</b>',
      :review    => '<!---ユーザーレビュー -->' #読者レビュー
    }
    mode = :none
    data = Hash.new()
    
    #----- read and sepalate by key(mode_tbl) ------
    str.each{|line|
      l =  line.chomp.chomp("\r").kconv(@insystemKcode)
      mode_tbl.each{ |key,keydata|
        mode = key if /#{keydata}/s =~ l
      }
      if mode != :none
        if data[mode]
          data[mode] << l
        else
          data[mode] = l
        end
      end
    }
    
    #----- parse -----
    if !data[:image].nil?
      if data[:image] =~ /#{mode_tbl[:image]}.+?(<td.+?)<\/td>/smi
        tmp = $1
        if tmp =~ /<img src=\"([^\"]+?)\"/
          @image = $1.strip
        end
      end
    end
    
    if !data[:publisher].nil?
      if data[:publisher] =~ /#{mode_tbl[:publisher]}.+?(<td.+?)<\/td>/smi
        @publisher = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:title].nil?
      if data[:title] =~ /#{mode_tbl[:title]}.+?(<td.+?)<\/td>/smi
        @title = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:label].nil?
      if data[:label] =~ /#{mode_tbl[:label]}.+?(<td.+?)<\/td>/smi
        @label = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:author].nil?
      if data[:author] =~ /#{mode_tbl[:author]}.+?(<td.+?)<\/td>/smi
        @author = $1.gsub(/<[^>]+>/,'').strip
        authors, translators = [], []
        @author.scan(/(.+)\/著/){ |author| authors << author }
        @author.scan(/([^\s]+)\/訳/){ |translator| translators << translator }
        @author = authors.join("\t") if authors.size > 0
        @translator = translators.join("\t") if translators.size > 0
      end
    end
    
    if !data[:pubdate].nil?
      if data[:pubdate] =~ /#{mode_tbl[:pubdate]}.+?(<td.+?)<\/td>/smi
        @pubdate = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:size].nil?
      if data[:size] =~ /#{mode_tbl[:size]}.+?(<td.+?)<\/td>/smi
        @pubsize = $1.gsub(/<[^>]+>/,'').strip
        if @pubsize =~ /^(.+P)(.+)$/
          @page, @pubsize = $1.strip, $2.strip
        end
      end
    end
    
    if !data[:isbn].nil?
      if data[:isbn] =~ /#{mode_tbl[:isbn]}.+?(<td.+?)<\/td>/smi
        @isbn = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:price].nil?
      if data[:price] =~ /#{mode_tbl[:price]}.+?(<td.+?)<\/td>/smi
        @price = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:stock].nil?
      if data[:stock] =~ /#{mode_tbl[:stock]}.+?(<td.+?)<\/td>/smi
        @stock = $1.gsub(/<[^>]+>/,'').strip
        if @stock =~ /^(.+)（(.+)）$/
          @stock, @delivery = $1.strip, $2.strip
        else
          @delivery = @stock
        end
      end
    end
    
    if !data[:desc].nil?
      if data[:desc] =~ /#{mode_tbl[:desc]}.+?(<td.+?)<\/td>/smi
        @desc = $1.gsub(/\s+/mi,' ').gsub(/<br>/mi,"\t").gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:index].nil?
      if data[:index] =~ /#{mode_tbl[:index]}.+?(<td.+?)<\/td>/smi
        @index = $1.gsub(/\s+/mi,' ').gsub(/<br>/mi,"\t").gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:user].nil?
      if data[:user] =~ /#{mode_tbl[:user]}.+?(<td.+?)<\/td>/smi
        @user = $1.gsub(/<[^>]+>/,'').strip
      end
    end
    
    if !data[:review].nil?
      if data[:review] =~ /読者の採点：.+?(（.+?）)/smi
        @review_point = $1.gsub(/<[^>]+>/,'').strip
      end
      if data[:review] =~ %r|<a href="(.+?)">すべてのレビューを見る \((.+?)\)</a>|mis
        url, @review_num = $1.strip, $2.strip
        @review_num.chomp!('件')
        if url =~ /sid=(\d+)/
          @reviews_id = $1.strip
        end
      end
      if @reviews_id.nil?
        if data[:review] =~ %r|<a href=".+?sid=(\d+).*?">レビューを書く</a>|mis
          @reviews_id = $1.strip
        end
      end
      
      reviewlist = []
      data[:review].scan(%r|<td><small>(.+?)</small></td>|mi){ |rev_str|
        rev_str = rev_str.to_s
        title, reviewer, date, value, comment = nil, nil, nil, nil, nil
        if rev_str =~ %r|<b>(.+?)</b>\s+投稿者：(.+?)　(.+?)<BR>(.+)|smi
          title, reviewer, date, comment = $1, $2, $3, $4
          title = title.gsub(/<[^>]+>/,'').strip
          reviewer = reviewer.gsub(/<[^>]+>/,'').strip
          date = date.gsub(/<[^>]+>/,'').strip
          value = 0
          comment.scan(/1s.gif/){ value += 1 }
          comment = comment.gsub(/<[^>]+>/,'').strip
          if comment =~ /^-\s*/
            comment = $'
          end
          reviewlist << [title, reviewer, date, value, comment]
        end
      }
      @reviewlist = reviewlist.flatten.join("\t") if reviewlist.size > 0
    end
    
    #読者レビューは別頁 "http://messages.yahoo.co.jp/bbs?.mm=&action=l&tid=#{id}&sid=#{reviews_id}"
    #店長レビューは別頁 "/esb_review/#{id}.html"
  end
  
  def set_info #===== set book-info(@info) from local-value =====
    super
    #----- set info add ----- #yahoo独自の情報類
    @info["user"]         = @user         #読者対象
    @info["review_point"] = @review_point #読者レビュー平均点
    @info["reviews_id"]   = @reviews_id   #レビュー制御用ID
    @info["reviewlist"]   = @reviewlist   #レビューリスト
#    @info["ownerreviewlist"] = @ownerreviewlist #店長レビューリスト
  end
end

if __FILE__ == $0
  p CacheBib_yahoo.new.get_data('4-15-010739-4')
#  p CacheBib_yahoo.new.get_data('4-10-610003-7')
#  p CacheBib_yahoo.new.get_data('4-594-04256-2')
end