#!/usr/bin/env ruby
# esbooks.rb  ver.0.4 2002.3.21-2003.1.20
# ISBN-search -> detail-parse
# original; isbn2bib.rb for bk1 by takahashi

#＞集英社コバルト文庫
#:genre     => '■ジャンル名',
#:delivery  => '■発送可能時期',
#売上ランク
#評価ポイント
#review
#以上、やり残し

require 'net/http'
require 'cgi'
#require 'simpleuri'
#require 'webagent'

class CacheBib_esbooks < CacheBib
  attr_accessor :isbn
  attr_accessor :flag_review #, :affiliate_code
  attr_accessor :url_site
  attr_accessor :title, :author, :publisher, :size, :pubdate, :price, :desc
  attr_accessor :image, :translator, :genre, :genre1, :genre2, :genre3
  attr_accessor :reader, :stock, :delivery
  attr_accessor :reviews, :reviewlist
  attr_accessor :insystemKcode, :outputKcode, :messageKcode

  attr_accessor :product_cd

  def initialize ()
    super
    @reviews = Array.new()
    @flag_review = true
    @url_site = 'http://www.esbooks.co.jp/' 
    @insystemKcode = 3 #sjis #1-jis,2-euc,3-sjis
    @outputKcode   = 2 #euc
    @messagetKcode = 3 #sjis
  end

#===== ===== parts ===== =====
  def word_delivery ()
    delivery = ''
    flag = false
    # title-check -> no-info / get-info check
    if !@title.nil?
      flag = true
      if @delivery.nil?
        delivery = "ない"
      else
        if @delivery =~ /翌日〜３日/
          delivery = "1〜3日"
        elsif @delivery =~ /３日〜６日/
          delivery = "3〜6日"
        elsif @delivery =~ /３日〜３週間/
          delivery = "3〜21日"
        else
          delivery = "？"
        end
      end
    else
      message = "WAR; maybe data-error, no-title."
      @stderr.puts message.kconv(@messageKcode)
    end
    return delivery, flag
  end

#===== ===== url ===== =====
  def url_isbnsearch ()
#    url = @url_site + "bks.svl?CID=BKS504&access_method=isbn_cd&input_data=#{@isbn}"
    url = @url_site + "search?isbn=#{@isbn}"
    return url
  end

  def url_bibinfo ()
#    url = @url_site + "bks.svl?CID=BKS503&product_cd=#{@product_cd}"
    url = @url_site + "product/keyword/keyword?accd=#{@product_cd}"
    return url
  end

#===== ===== parse ===== =====
  def parse_isbnsearch (str)
    count = 0
    str.each{ |line|
#      if line =~ %r|\s*<a href=/bks.svl\?.*&product_cd=(\d+)&.*>(.*?)</a></TD>| then
       if line =~ %r|<a href="/product/keyword/keyword\?accd=(\d+)">(.+?)</a>|s then
        @product_cd, @title = $1, $2
        if @product_cd.nil?
          @title = nil
        else
          count = count + 1
        end
      end
    }
    if count > 1
      message = "WAR; ISBN-search hit multi-column. ISBN; " + @isbn
      @stderr.puts message.kconv(@messageKcode)
    end
  end

  def parse_bibinfo (str)
    #----- set key and data ------
    mode_tbl = {
      :title     => 'class="colum-title"',
      :image     => '/bks/images/i',
      :author    => '著者／訳者名',
      :publisher => '出版社名',
      :genre     => '■ジャンル名',
      :pubdate   => '発行年月',
      :isbn      => 'ISBNコード',
      :size      => 'ページ・サイズ',
      :reader    => '読者対象',
      :stock     => '在庫状況',
      :price     => '販売価格',
      :delivery  => '■発送可能時期',
      :content   => '【本の内容】',
      :content2  => '【出版社からのコメント】',
      :review    => '<a href="/bks.svl\?start&CID=BKS401&shop_cd=1&cust_id=', 
      :review2   => '<a href="javascript:OpenWindow\(\'/bks.svl\?start&CID=BKS420&shop_cd=1&cust_id='
    }
#売上ランク
#評価ポイント
    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[:title].nil?
      if data[:title] =~ /^(.+?)<\/strong>/i
        @title = $1.gsub(/<[^>]+>/,'').strip
      end
    end

    data[:image] =~ Regexp.new("<img src=\"(.*?)\"")
    image = $1
    if !image.nil?
      @image = @url_site.chomp('/') + image if image !~ /books_icn.gif/
    end

    if !data[:author].nil?
#      data[:author] =~ Regexp.new("<a href=.*?&input_data=(.*?)&access_method")
#      string = $1.to_s.kconv(@insystemKcode)
      data[:author] =~ Regexp.new("：(.*)")
      string = $1.gsub(/<[^>]+>/,'').strip if !$1.nil?
      @author = string.strip if !string.nil?
      if string =~  /(.*)／(〔ほか〕)?〔?[著|編]〕?/s
        names = $1.split(/〔?[著|編]〕?[\s|　|$]/s)
        names.collect! { |name| name.strip.chomp('／') }
        @author = names.join("\t")
      end
      if string =~  /.*／(〔ほか〕)?〔?[著|編]〕?(.*)／(〔ほか〕)?〔?訳〕?/s
        names = $2.split( /〔?訳〕?[\s|　|$]/s)
        names.collect! { |name| name.strip.chomp('／') }
        names.collect! { |name| name.gsub(/^　*/,'').strip }
        @translator = names.join("\t")
      end
    end

    data[:publisher] =~ Regexp.new("：(.*)")
    @publisher = $1.gsub(/<[^>]+>/,'').strip if !$1.nil?

    data[:genre] =~ Regexp.new("：(.*)")
    @genre = $1.gsub(/<[^>]+>/,'').strip if !$1.nil?
    data[:genre] =~ Regexp.new("genre1_name=(.+?)&")
    @genre1 = $1
    data[:genre] =~ Regexp.new("genre2_name=(.+?)&")
    @genre2 = $1
    data[:genre] =~ Regexp.new("genre3_name=(.+?)&")
    @genre3 = $1

    data[:pubdate] =~ Regexp.new("：(.*)")
    @pubdate = $1.gsub(/<[^>]+>/,'').strip if !$1.nil?

    data[:isbn] =~ Regexp.new("：(.*)[　]?")
    @isbn = $1.gsub(/<[^>]+>/,'').delete('　').strip if !$1.nil?

    data[:size] =~ Regexp.new("：(.*)")
    @size = $1.gsub(/<[^>]+>/,'').strip if !$1.nil?

    data[:reader] =~ Regexp.new("：(.*)")
    @reader = $1.gsub(/<[^>]+>/,'').strip if !$1.nil?

    data[:stock] =~ Regexp.new("：(.*)")
    @stock = $1.strip if !$1.nil?
    if @stock =~ /(.+?)<\/tr>/i
      @stock = $1.gsub(/<[^>]+>/,'').strip
    end

    data[:price] =~ Regexp.new("：(.*?)</tr>")
    str = $1
    @price = str.gsub(/<[^>]+?>/,'').strip.delete("\s") if !str.nil?

#    if !data[:delivery].nil?
#      data[:delivery] =~ Regexp.new("：(.*?)&gt;")
#      @delivery = $1.to_s.gsub(/<[^>]+>/,'').strip
#    end
    if @stock =~ /（(.+?)）/s
      @delivery = $1
    end

    data[:content] =~ Regexp.new("#{mode_tbl[:content]}(.*?)</table>")
    @desc = $1.to_s.gsub(/<[^>]+>/,'').strip
    @desc = '' if @desc == @delivery

    data[:content2] =~ Regexp.new("#{mode_tbl[:content2]}(.*?)</table>")
    tmp = $1.to_s.gsub(/<[^>]+>/,'').strip
    tmp = '' if tmp == @delivery
    @desc += tmp

    reviewsplit = data[:review].to_s.split(mode_tbl[:review])
    reviewlist = []
    reviewsplit.each { |column|
      if column =~ /cust_id=(.+?)'/
        custid = $1
        reviewlist << custid
      end
      if column =~ /<strong>(.+?)店長/
        custid = $1.strip
        reviewlist << custid
      end
    }
    reviewsplit = data[:review2].to_s.split(mode_tbl[:review2])
    reviewsplit.each { |column|
      if column =~ /cust_id=(.+?)'/
        custid = $1
        reviewlist << custid
      end
      if column =~ /<strong>(.+?)店長/
        custid = $1.strip
        reviewlist << custid
      end
    }
    @reviewlist = reviewlist.join("\t") if reviewlist.size > 0

    if @flag_review
      reviewsplit = data[:review].to_s.split(mode_tbl[:review])
      reviewsplit.each{ |column|
        review = {}
        if column =~ /cust_id=(.+?)'/
          review["cust_id"] = $1
        end
        if column =~ /<strong>(.+?)店長の『(.+?)』コーナー/s
          review["reviewer"] = $1.strip
          review["corner"]   = $2.strip
        end
        if column =~ /#{review["reviewer"]}\s*\[(.+?)\]\s*−\s*(.+?日)(.+?)<\/td>/s
          review["value"] = $1
          review["date"]  = $2
          review["content"]  = $3
          review["content"].gsub!(/^<.+?>/,'').strip!
          review["content"].gsub!(/\n/,"\t")
        end
        @reviews << review if review.size > 0
      }
      reviewsplit = data[:review2].to_s.split(mode_tbl[:review2])
      reviewsplit.each{ |column|
        review = {}
        if column =~ /cust_id=(.+?)'/
          review["cust_id"] = $1
        end
        if column =~ /<strong>(.+?)店長/s
          review["reviewer"] = $1.strip
        end
        if column =~ /<\/strong>(.+?)<br>(.+?)<\/td>/s
          review["content"] = $1
          review["date"]  = $2
          review["content"].gsub!(/<[^>]+>/,'').gsub!(/\s+/,'').strip!
          review["date"].strip!
        end
        @reviews << review if review.size > 0
      }
    end
  end

#===== ===== main ===== =====
  def set_info #===== set local-value from @info (from cache) =====
    #----- set book-info (from cache) -----
    @flag_overprint  = true if @info["overprint"]
    @product_cd       = @info["product_cd"]
    @flag_isbnsearch = true if !@product_cd

    @title    = @info["title"]
    @delivery = @info["delivery"]
  end

  def get_info #===== get book-info(@info) from url =====
    message = 'CHK; access; ' + @url_site + ' -> ' + @isbn
    @stderr.puts message.kconv(@messageKcode)
    #----- get detail-code -----
    if @flag_isbnsearch
      agent = WebAgent.new()
      agent.uri = url_isbnsearch
      agent.get()
      body = agent.body.kconv(@insystemKcode)
      parse_isbnsearch(body)
    end

    #----- get book-info -----
    agent = WebAgent.new()
    agent.uri = url_bibinfo
    agent.get()
    body = agent.body.kconv(@insystemKcode)
    parse_bibinfo(body)

    if !@flag_overprint
      @info = {}
    end
    @info["isbn"]       = @isbn
    @info["title"]      = @title
    @info["author"]     = @author
    @info["publisher"]  = @publisher
    @info["size"]       = @size
    @info["pubdate"]    = @pubdate
    @info["price"]      = @price
    @info["desc"]       = @desc
    @info["image"]      = @image
    @info["translator"] = @translator
    @info["genre"]      = @genre
    @info["genre1"]     = @genre1
    @info["genre2"]     = @genre2
    @info["genre3"]     = @genre3

    @info["reader"]     = @reader
    @info["stock"]      = @stock
    @info["delivery"]   = @delivery

    @info["product_cd"]  = @product_cd
    @info["reviewlist"] = @reviewlist

    @info["overprint"]  = true if @flag_overprint

    if @flag_review
      @info["reviews"] = @reviews
    end
  end
end