#!/usr/bin/env ruby
# webcat.rb(books)  ver.0.1 2003.1.27-2003.1.27
# ISBN-search -> detail-parse
# original; isbn2bib.rb for bk1 by takahashi

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

class CacheBib_webcat < CacheBib
  attr_accessor :isbn
  attr_accessor :flag_review #, :affiliate_code
  attr_accessor :url_site
  attr_accessor :title, :title2, :title3, :author, :author2, :publisher
  attr_accessor :size, :note, :note2, :type, :type2, :code, :case
  attr_accessor :desc, :index
  attr_accessor :ncid, :accessid
  attr_accessor :reviews, :delivery
  attr_accessor :insystemKcode, :outputKcode, :messageKcode

  attr_accessor :detail_id

  def initialize ()
    super
    @reviews = Array.new()
    @flag_review = true
    @url_site = 'http://webcatplus.nii.ac.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 + "assoc.cgi?mode=equal0&isbn=#{@isbn}"
    return url
  end

  def url_bibinfo ()
    url = @url_site + "tosho.cgi?mode=tosho&#{@accessid}=#{@ncid}"
    return url
  end

#===== ===== parse ===== =====
  def parse_isbnsearch (str)
    count = 0
    str.each{ |line|
      if line =~ %r|<a href="tosho.cgi\?mode=tosho&(..ID)\=([\w\d]+)"[^>]+?><b>(.*?)</b></a>| then
#      if line =~ %r|<a href="tosho.cgi\?mode=tosho&(.+?)"><b>(.*?)</b></a>| then
        @accessid, @ncid, @title = $1, $2, $3
        if @ncid.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 = {
#      :data1     => '<div style="margin-top:11px;">',
      :desc      => '<img src="/webcatplus/widgets/images/etc/b01_naiyou.gif" border="0" width="46" height="13" alt="内容" title="内容" />',
      :index     => '<img src="/webcatplus/widgets/images/etc/b01_mokuji.gif" border="0" width="46" height="14" alt="目次" title="目次" />',
#      :data2     => '<img src="/webcatplus/widgets/images/etc/b01_jyouho.gif" border="0" width="54" height="13" alt="書誌情報" title="書誌情報" />',
      :title     => '<font size="2" class="fs100">タイトル</font>',
      :author    => '<font size="2" class="fs100">責任表示</font>',
      :type      => '<font size="2" class="fs100">資料種別</font>',
      :type2     => '<font size="2" class="fs100">版表示</font>',
      :publisher => '<font size="2" class="fs100">出版事項</font>',
      :size      => '<font size="2" class="fs100">形態事項</font>',
      :isbn      => '<font size="2" class="fs100">ISBN</font>', #dummy
      :note2     => '<font size="2" class="fs100">内容著作注記</font>',
      :note      => '<font size="2" class="fs100">注記</font>',
      :title2    => '<font size="2" class="fs100">別タイトル</font>',
      :title3    => '<font size="2" class="fs100">統一タイトル</font>',
      :author2   => '<font size="2" class="fs100">著者標目</font>',
      :code      => '<font size="2" class="fs100">分類</font>',
      :case      => '<font size="2" class="fs100">件名</font>'
    }
    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 -----
    keys = [:desc, :index]
    stocks = [nil, nil]
    keys.each_index{ |i|
      key = keys[i]
      if !data[key].nil?
        if data[key] =~ /#{mode_tbl[key]}.+?<td>(.+?)<\/td>/smi
          stocks[i] = $1.gsub(/\s+/, ' ').gsub(/<br>/i, "\t").gsub(/<[^>]+>/,'').strip
        end
      end
    }
    @desc, @index = stocks

    keys = [:title, :author, :type, :type2, :publisher, :size, :note2, :note, :title2, :title3, :author2, :code, :case]
    stocks = [nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil, nil]
    keys.each_index{ |i|
      key = keys[i]
      if !data[key].nil?
        if data[key] =~ /#{mode_tbl[key]}.+?<td>(.+?)<\/td>/smi
          stocks[i] = $1.gsub(/\s+/, ' ').gsub(/<br>/i, "\t").gsub(/<[^>]+>/,'').strip
        end
      end
    }
    @title, @author, @type, @type2, @publisher, @size, @note2, @note, @title2, @title3, @author2, @code, @case = stocks
  end

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

    @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["desc"]       = @desc
    @info["index"]      = @index
    @info["author"]     = @author
    @info["type"]       = @type
    @info["type2"]      = @type2
    @info["publisher"]  = @publisher
    @info["size"]       = @size
    @info["note2"]      = @note2
    @info["note"]       = @note
    @info["title2"]     = @title2
    @info["title3"]     = @title3
    @info["pubdate"]    = @pubdate
    @info["author2"]    = @author2
    @info["code"]       = @code
    @info["case"]       = @case

    @info["ncid"]       = @ncid
    @info["accessid"]   = @accessid

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

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