cbeta 0.0.1

data/lib/cbeta/gaiji.rb

@@ -0,0 +1,39 @@
+require 'json'
+
+# 存取 CBETA 缺字資料庫
+class CBETA::Gaiji
+	# 載入 CBETA 缺字資料庫
+  def initialize()
+    fn = File.join(File.dirname(__FILE__), 'gaiji.json')
+    @gaijis = JSON.parse(File.read(fn))
+  end
+
+  # 傳入缺字 CB 碼，傳回 hash 缺字資訊
+  #
+  # 例如：
+  #
+  #   g = Cbeta::Gaiji.new
+  #   g["CB01002"]
+  #
+  # 回傳：
+  #   {
+  #     "zzs": "[得-彳]",
+  #     "unicode": "3775",
+  #     "unicode-char": "㝵",
+  #     "zhuyin": [ "ㄉㄜˊ", "ㄞˋ" ]
+  #   }
+  def [](cb)
+  	@gaijis[cb]
+  end
+
+  # 傳入缺字 CB 碼，傳回注音 array
+  #
+  # 例如：
+  #
+  #   g = Cbeta::Gaiji.new
+  #   g.zhuyin("CB00023") # return [ "ㄍㄢˇ", "ㄍㄢ", "ㄧㄤˊ", "ㄇㄧˇ", "ㄇㄧㄝ", "ㄒㄧㄤˊ" ]
+  def zhuyin(cb)
+  	return nil unless @gaijis.key? cb
+    @gaijis[cb]['zhuyin']
+  end
+end

data/lib/cbeta/html_to_text.rb

@@ -0,0 +1,151 @@
+require 'fileutils'
+require 'nokogiri'
+
+# 將 CBETA HTML 轉為 純文字(含行首資訊)
+#
+# example:
+#
+#   h2t = CBETA::HTMLToText.new('/temp/cbeta-html', '/temp/cbeta-text')
+#   h2t.convert("T01")  # 轉換大正藏第一冊
+class CBETA::HTMLToText
+  # html_root:: 來源 HTML 路徑
+  # out_root:: 輸出路徑
+  def initialize(html_root, out_root)
+    @html_root = html_root
+    @out_root = out_root
+  end
+
+  # 例如執行大正藏第一冊：
+  #
+  # convert("T01")
+  def convert(arg)
+    @dirty = false
+    @vol = arg.upcase
+    @corpus = @vol[0]
+    handle_vol
+  end
+
+  private
+
+  def traverse(e)
+    r = ''
+    e.children.each { |c|
+      r += handle_node(c)
+    }
+    r.gsub('　', '')
+  end
+
+  def handle_text(e)
+    s = e.content().chomp
+    return '' if s.empty?
+    s.gsub(/[\n，、—！。：「]/, '')
+  end
+
+  def handle_span(e)
+    r = ''
+    case e['class']
+    when 'doube-line-note'
+      r = traverse(e)
+      unless r.start_with? '（'
+        r = "(#{r})"
+      end
+    when 'lb'
+      if @dirty
+        r += "\n"
+      else
+        @dirty = true
+      end
+      # 行首資訊 T05n0220a 改為 T05n0220
+      lb = e['id'].sub(/^(T0\dn0220)[a-z](.*)$/, '\1\2')
+      r += lb + '║'
+    when 'lineInfo'
+    when 'ranja'
+      r = '【◇】'
+    when 'siddam'
+      r = '【◇】'
+    when 'star'
+    else
+      r = traverse(e)
+    end
+    r
+  end
+
+  def handle_node(e)
+    return '' if e.comment?
+    return handle_text(e) if e.text?
+    r = ''
+    case e.name
+    when 'a'
+      if e['class'] == 'gaijiAnchor'
+        id = e['href'][1..-1]
+        r = @gaiji[id]
+      else
+        r = traverse(e)
+      end
+    when 'div'
+      if e['id'] != 'back'
+        r = traverse(e)
+      end
+    when 'head'
+    when 'p'
+      if e['class'] == 'figure'
+        r = '【圖】'
+      else
+        r = traverse(e)
+      end
+    when 'span'
+      r = handle_span(e)
+    else
+      r = traverse(e)
+    end
+    r
+  end
+
+  def prepare_folder()
+    folder = File.join(@out_root, @corpus, @vol)
+    FileUtils.remove_dir(folder, force=true)
+    FileUtils.mkdir_p(folder)
+    folder
+  end
+
+  def handle_file(path)
+    sutra = File.basename(path, ".*")
+    sutra.sub!(/^(.*)_.*$/, '\1')
+    sutra.sub!(/(T\d\dn0220).*$/, '\1') # T0220 BM 沒有分 a, b, c...
+
+    if sutra != @last_sutra
+      txt_fn = sutra + '.txt'
+      txt_path = File.join(@folder_out, txt_fn)
+      puts "h2t #{txt_path}"
+      @fo = File.open(txt_path, 'w')
+      @last_sutra = sutra
+      @dirty = false
+    end
+
+    f = File.open(path)
+    doc = Nokogiri::HTML(f)
+    f.close
+
+    @gaiji = {}
+    doc.css("span.gaijiInfo").each { |e|
+      @gaiji[e['id']] = e['zzs']
+    }
+
+    text = traverse(doc.root)
+
+    # 悉曇字
+    text.gsub!(/(\((【◇】)+\)|（【◇】）|【◇】)+/, '【◇】')
+
+    @fo.write(text)
+  end
+
+  def handle_vol()
+    folder_in = File.join(@html_root, @corpus, @vol)
+    @folder_out = prepare_folder
+    @last_sutra = ''
+    Dir["#{folder_in}/*"].each { |f|
+      handle_file(f)
+    }
+  end
+
+end

data/lib/cbeta/p5a_to_html.rb

@@ -0,0 +1,664 @@
+require 'cgi'
+require 'date'
+require 'fileutils'
+require 'json'
+require 'nokogiri'
+require 'set'
+
+# 內容不輸出的元素
+PASS=['back', 'teiHeader']
+
+# 某版用字缺的符號
+MISSING = '－'
+
+# 處理 CBETA XML P5a
+#
+# CBETA XML P5a 可由此取得: https://github.com/cbeta-git/xml-p5a
+#
+# 轉檔規則請參考: http://wiki.ddbc.edu.tw/pages/CBETA_XML_P5a_轉_HTML
+class CBETA::P5aToHTML
+
+  # xml_root:: 來源 CBETA XML P5a 路徑
+  # out_root:: 輸出 HTML 路徑
+  def initialize(xml_root, out_root)
+    @xml_root = xml_root
+    @out_root = out_root
+    @gaijis = CBETA::Gaiji.new
+
+    # 載入 unicode 1.1 字集列表
+    fn = File.join(File.dirname(__FILE__), 'unicode-1.1.json')
+    json = File.read(fn)
+    @unicode1 = JSON.parse(json)
+  end
+
+  # 將 CBETA XML P5a 轉為 HTML
+  #
+  # 例如 轉出大正藏第一冊
+  #
+  #   x2h = CBETA::P5aToHTML.new('/PATH/TO/CBETA/XML/P5a', '/OUTPUT/FOLDER')
+  #   x2h.convert('T01')
+  #
+  # 例如 轉出大正藏全部
+  #
+  #   x2h = CBETA::P5aToHTML.new('/PATH/TO/CBETA/XML/P5a', '/OUTPUT/FOLDER')
+  #   x2h.convert('T')
+  #
+  # T 是大正藏的 ID, CBETA 的藏經 ID 系統請參考: http://www.cbeta.org/format/id.php
+  def convert(arg=nil)
+    return convert_all if arg.nil?
+
+    arg.upcase!
+    if arg.size == 1
+      handle_collection(arg)
+    else
+      if arg.include? '..'
+        arg.match(/^([^\.]+?)\.\.([^\.]+)$/) {
+          handle_vols($1, $2)
+        }
+      else
+        handle_vol(arg)
+      end
+    end
+  end
+
+  private
+
+  def convert_all
+    Dir.foreach(@xml_root) { |c|
+      next unless c.match(/^[A-Z]$/)
+      handle_collection(c)
+    }
+  end
+
+  def handle_anchor(e)
+    id = e['id']
+    if e.has_attribute?('id')
+      if id.start_with?('nkr_note_orig')
+        note = @notes[id]
+        note_text = traverse(note)
+        n = id[/^nkr_note_orig_(.*)$/, 1]
+        @back[@juan] += "<span class='footnote' id='n#{n}'>#{note_text}</span>\n"
+        return "<a class='noteAnchor' href='#n#{n}'></a>"
+      elsif id.start_with? 'fx'
+        return "<span class='star'>[＊]</span>"
+      end
+    end
+
+    if e.has_attribute?('type')
+      if e['type'] == 'circle'
+        return '◎'
+      end
+    end
+
+    ''
+  end
+
+  def handle_app(e)
+    r = ''
+    if e['type'] == 'star'
+      c = e['corresp'][1..-1]
+      r = "<a class='noteAnchor star' href='#n#{c}'></a>"
+    end
+    r + traverse(e)
+  end
+
+  def handle_byline(e)
+    r = '<p class="byline">'
+    r += "<span class='lineInfo'>#{@lb}</span>"
+    r += traverse(e)
+    r + '</p>'
+  end
+
+  def handle_collection(c)
+    @series = c
+    puts 'handle_collection ' + c
+    folder = File.join(@xml_root, @series)
+    Dir.foreach(folder) { |vol|
+      next if ['.', '..', '.DS_Store'].include? vol
+      handle_vol(vol)
+    }
+  end
+
+  def handle_corr(e)
+    r = ''
+    if e.parent.name == 'choice'
+      sic = e.parent.at_xpath('sic')
+      unless sic.nil?
+        @dila_note += 1
+        r = "<a class='noteAnchor dila' href='#dila_note#{@dila_note}'></a>"
+
+        note = @orig
+        sic_text = traverse(sic, 'back')
+        if sic_text.empty?
+          note += MISSING
+        else
+          note += sic_text
+        end
+        @back[@juan] += "<span class='footnote_dila' id='dila_note#{@dila_note}'>#{note}</span>\n"
+      end
+    end
+    r + "<span class='cbeta'>%s</span>" % traverse(e)
+  end
+
+  def handle_div(e)
+    @div_count += 1
+    n = @div_count
+    if e.has_attribute? 'type'
+      @open_divs << e
+      r = traverse(e)
+      @open_divs.pop
+      return "<!-- begin div#{n}--><div class='div-#{e['type']}'>#{r}</div><!-- end of div#{n} -->"
+    else
+      return traverse(e)
+    end
+  end
+
+  def handle_figure(e)
+    "<p class='figure'>%s</p>" % traverse(e)
+  end
+
+  def handle_g(e, mode)
+    # if 有 <mapping type="unicode">
+    #   if 在 unicode 1.1 範圍裡
+    #     直接採用
+    #   else
+    #     預設呈現 unicode, 但仍包缺字資訊，供點選開 popup
+    # else if 有 <mapping type="normal_unicode">
+    #   預設呈現 normal_unicode, 但仍包缺字資訊，供點選開 popup
+    # else if 有 normalized form
+    #   預設呈現 normalized form, 但仍包缺字資訊，供點選開 popup
+    # else
+    #   預設呈現組字式, 但仍包缺字資訊，供點選開 popup
+    gid = e['ref'][1..-1]
+    g = @gaijis[gid]
+    abort "Line:#{__LINE__} 無缺字資料:#{gid}" if g.nil?
+    zzs = g['zzs']
+    
+    if mode == 'txt'
+      return g['roman'] if gid.start_with?('SD')
+      if zzs.nil?
+        abort "缺組字式：#{g}"
+      else
+        return zzs
+      end
+    end
+
+    @char_count += 1
+
+    if gid.start_with?('SD')
+      case gid
+      when 'SD-E35A'
+        return '（'
+      when 'SD-E35B'
+        return '）'
+      else
+        return "<span class='siddam' roman='#{g['roman']}' code='#{gid}' char='#{g['sd-char']}'/>"
+      end
+    end
+    
+    if gid.start_with?('RJ')
+      return "<span class='ranja' roman='#{g['roman']}' code='#{gid}' char='#{g['rj-char']}'/>"
+    end
+   
+    default = ''
+    if g.has_key?('unicode')
+      if @unicode1.include?(g['unicode'])
+        return g['unicode-char'] # unicode 1.1 直接用
+      else
+        default = g['unicode-char']
+      end
+    end
+
+    nor = ''
+    if g.has_key?('normal_unicode')
+      nor = g['normal_unicode']
+      default = nor if default.empty?
+    end
+
+    if g.has_key?('normal')
+      nor += ', ' unless nor==''
+      nor += g['normal']
+      default = g['normal'] if default.empty?
+    end
+
+    default = zzs if default.empty?
+    
+    href = 'http://dict.cbeta.org/dict_word/gaiji-cb/%s/%s.gif' % [gid[2, 2], gid]
+    unless @back[@juan].include?(href)
+      @back[@juan] += "<span id='#{gid}' class='gaijiInfo' figure_url='#{href}' zzs='#{zzs}' nor='#{nor}'>#{default}</span>\n"
+    end
+    "<a class='gaijiAnchor' href='##{gid}'>#{default}</a>"
+  end
+
+  def handle_graphic(e)
+    url = File.basename(e['url'])
+    "<span imgsrc='#{url}' class='graphic'></span>"
+  end
+
+  def handle_head(e)
+    r = ''
+    unless e['type'] == 'added'
+      r = "<p class='head'>%s</p>" % traverse(e)
+    end
+    r
+  end
+
+  def handle_item(e)
+    "<li>%s</li>\n" % traverse(e)
+  end
+
+  def handle_juan(e)
+    "<p class='juan'>%s</p>" % traverse(e)
+  end
+
+  def handle_l(e)
+    if @lg_type == 'abnormal'
+      return traverse(e)
+    end
+
+    @in_l = true
+
+    doc = Nokogiri::XML::Document.new
+    cell = doc.create_element('div')
+    cell['class'] = 'lg-cell'
+    cell.inner_html = traverse(e)
+    
+    if @first_l
+      parent = e.parent()
+      if parent.has_attribute?('rend')
+        indent = parent['rend'].scan(/text-indent:[^:]*/)
+        unless indent.empty?
+          cell['style'] = indent[0]
+        end
+      end
+      @first_l = false
+    end
+    r = cell.to_s
+    
+    unless @lg_row_open
+      r = "\n<div class='lg-row'>" + r
+      @lg_row_open = true
+    end
+    @in_l = false
+    r
+  end
+
+  def handle_lb(e)
+    @char_count = 1
+    @lb = e['n']
+    line_head = @sutra_no + '_p' + e['n']
+    r = ''
+    #if e.parent.name == 'lg' and $lg_row_open
+    if @lg_row_open && !@in_l
+      # 每行偈頌放在一個 lg-row 裡面
+      # T46n1937, p. 914a01, l 包雙行夾註跨行
+      # T20n1092, 337c16, lb 在 l 中間，不結束 lg-row
+      r += "</div><!-- end of lg-row -->"
+      @lg_row_open = false
+    end
+    r + "<span class='lb' \nid='#{line_head}'>#{line_head}</span>"
+  end
+
+  def handle_lem(e)
+    r = ''
+    w = e['wit']
+    if w.include? 'CBETA' and not w.include? @orig
+      @dila_note += 1
+      r = "<a class='noteAnchor dila' href='#dila_note#{@dila_note}'></a>"
+      r += "<span class='cbeta'>%s</span>" % traverse(e)
+
+      note = lem_note_cf(e)
+      note += lem_note_rdg(e)
+      @back[@juan] += "<span class='footnote_dila' id='dila_note#{@dila_note}'>#{note}</span>\n"
+    else
+      r = traverse(e)
+    end
+    r
+  end
+
+  def handle_lg(e)
+    r = ''
+    @lg_type = e['type']
+    if @lg_type == 'abnormal'
+      r = "<p class='lg-abnormal'>" + traverse(e) + "</p>"
+    else
+      @first_l = true
+      doc = Nokogiri::XML::Document.new
+      node = doc.create_element('div')
+      node['class'] = 'lg'
+      if e.has_attribute?('rend')
+        rend = e['rend'].gsub(/text-indent:[^:]*/, '')
+        node['style'] = rend
+      end
+      @lg_row_open = false
+      node.inner_html = traverse(e)
+      if @lg_row_open
+        node.inner_html += '</div><!-- end of lg -->'
+        @lg_row_open = false
+      end
+      r = "\n" + node.to_s
+    end
+    r
+  end
+
+  def handle_list(e)
+    "<ul>%s</ul>" % traverse(e)
+  end
+
+  def handle_milestone(e)
+    r = ''
+    if e['unit'] == 'juan'
+
+      r += "</div>" * @open_divs.size  # 如果有 div 跨卷，要先結束, ex: T55n2154, p. 680a29, 跨 19, 20 兩卷
+      @juan = e['n'].to_i
+      @back[@juan] = @back[0]
+      r += "<juan #{@juan}>"
+      @open_divs.each { |d|
+        r += "<div class='#{d['type']}'>"
+      }
+    end
+    r
+  end
+
+  def handle_mulu(e)
+    r = ''
+    if e['type'] == '品'
+      @pass << false
+      r = "<mulu class='pin' s='%s'/>" % traverse(e, 'txt')
+      @pass.pop
+    end
+    r
+  end
+
+  def handle_node(e, mode)
+    return '' if e.comment?
+    return handle_text(e, mode) if e.text?
+    return '' if PASS.include?(e.name)
+    r = case e.name
+    when 'anchor'    then handle_anchor(e)
+    when 'app'       then handle_app(e)
+    when 'byline'    then handle_byline(e)
+    when 'corr'      then handle_corr(e)
+    when 'div'       then handle_div(e)
+    when 'figure'    then handle_figure(e)
+    when 'foreign'   then ''
+    when 'g'         then handle_g(e, mode)
+    when 'graphic'   then handle_graphic(e)
+    when 'head'      then handle_head(e)
+    when 'item'      then handle_item(e)
+    when 'juan'      then handle_juan(e)
+    when 'l'         then handle_l(e)
+    when 'lb'        then handle_lb(e)
+    when 'lem'       then handle_lem(e)
+    when 'lg'        then handle_lg(e)
+    when 'list'      then handle_list(e)
+    when 'mulu'      then handle_mulu(e)
+    when 'note'      then handle_note(e)
+    when 'milestone' then handle_milestone(e)
+    when 'p'         then handle_p(e)
+    when 'rdg'       then ''
+    when 'reg'       then ''
+    when 'sic'       then ''
+    when 'sg'        then handle_sg(e)
+    when 't'         then handle_t(e)
+    else traverse(e)
+    end
+    r
+  end
+
+
+  def handle_note(e)
+    n = e['n']
+    if e.has_attribute?('type')
+      t = e['type']
+      case t
+      when 'equivalent'
+        return ''
+      when 'orig'
+        @pass << false
+        s = traverse(e)
+        @pass.pop
+        @back[@juan] += "<span class='footnote_orig' id='n#{n}'>#{s}</span>\n"
+
+        if @mod_notes.include? n
+          return ''
+        else
+          return "<a class='noteAnchor' href='#n#{n}'></a>"
+        end
+      when 'mod'
+        @pass << false
+        s = traverse(e)
+        @pass.pop
+        @back[@juan] += "<span class='footnote_cb' id='n#{n}'>#{s}</span>\n"
+        return "<a class='noteAnchor' href='#n#{n}'></a>"
+      when 'rest'
+        return ''
+      else
+        return '' if t.start_with?('cf')
+      end
+    end
+
+    if e.has_attribute?('resp')
+      return '' if e['resp'].start_with? 'CBETA'
+    end
+
+    if e.has_attribute?('place') && e['place']=='inline'
+      r = traverse(e)
+      return "<span class='doube-line-note'>#{r}</span>"
+    else
+      return traverse(e)
+    end
+  end
+
+  def handle_p(e)
+    r = '<p>'
+    r += "<span class='lineInfo'>#{@lb}</span>"
+    r += traverse(e)
+    r + '</p>'
+  end
+
+  def handle_sg(e)
+    '(' + traverse(e) + ')'
+  end
+
+  def handle_sutra(xml_fn)
+    puts "handle sutra #{xml_fn}"
+    @back = { 0 => '' }
+    @char_count = 1
+    @dila_note = 0
+    @div_count = 0
+    @in_l = false
+    @juan = 0
+    @lg_row_open = false
+    @mod_notes = Set.new
+    @open_divs = []
+    @sutra_no = File.basename(xml_fn, ".xml")
+
+    text = parse_xml(xml_fn)
+
+    # 註標移到 lg-cell 裡面，不然以 table 呈現 lg 會有問題
+    text.gsub!(/(<a class='noteAnchor'[^>]*><\/a>)(<div class="lg-cell"[^>]*>)/, '\2\1')
+    
+    juans = text.split(/(<juan \d+>)/)
+    open = false
+    fo = nil
+    juan_no = nil
+    fn = ''
+    buf = ''
+    # 一卷一檔
+    juans.each { |j|
+      if j =~ /<juan (\d+)>$/
+        juan_no = $1.to_i
+        fn = "#{@sutra_no}_%03d.htm" % juan_no
+        output_path = File.join(@out_folder, fn)
+        fo = File.open(output_path, 'w')
+        open = true
+        s = <<eos
+<html>
+<head>
+  <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+  <meta name="filename" content="#{fn}" />
+  <title>#{@title}</title>
+</head>
+<body>
+<!-- 
+  來源 XML CBETA P5a: https://github.com/cbeta-org/xml-p5a.git
+  轉檔程式: Dropbox/DILA-DA/cbeta-html/bin/x2h.rb Version #{Date.today}
+  說明文件: http://wiki.ddbc.edu.tw/pages/CBETA_XML_P5a_%E8%BD%89_HTML
+-->
+<div id='body'>
+eos
+        fo.write(s)
+        fo.write(buf)
+        buf = ''
+      elsif open
+        fo.write(j + "\n</div><!-- end of div[@id='body'] -->\n")
+        fo.write("<div id='back'>\n" + @back[juan_no] + "</div>\n")
+        fo.write('</body></html>')
+        fo.close
+      else
+        buf = j
+      end
+    }
+  end
+
+  def handle_t(e)
+    if e.has_attribute? 'place'
+      return '' if e['place'].include? 'foot'
+    end
+    traverse(e)
+  end
+
+  def handle_text(e, mode)
+    s = e.content().chomp
+    return '' if s.empty?
+    return '' if e.parent.name == 'app'
+
+    # cbeta xml 文字之間會有多餘的換行
+    r = s.gsub(/[\n\r]/, '')
+
+    # 把 & 轉為 &amp;
+    r = CGI.escapeHTML(r)
+
+    # 正文區的文字外面要包 span
+    if @pass.last and mode=='html'
+      r = "<span class='t' l='#{@lb}' w='#{@char_count}'>#{r}</span>"
+      @char_count += r.size
+    end
+    r
+  end
+
+  def handle_vol(vol)
+    puts 'x2h ' + vol
+    if vol.start_with? 'T'
+      @orig = "【大】"
+    else
+      abort "未處理底本"
+    end
+    @vol = vol
+    @series = vol[0]
+    @out_folder = File.join(@out_root, @series, vol)
+    FileUtils.remove_dir(@out_folder, force=true)
+    FileUtils::mkdir_p @out_folder
+    
+    source = File.join(@xml_root, @series, vol)
+    Dir[source+"/*"].each { |f|
+      handle_sutra(f)
+    }
+  end
+
+  def handle_vols(v1, v2)
+    @series = v1[0]
+    folder = File.join(IN, @series)
+    Dir.foreach(folder) { |vol|
+      next if vol < v1
+      next if vol > v2
+      handle_vol(vol)
+    }
+  end
+
+  def lem_note_cf(e)
+    # ex: T32n1670A.xml, p. 703a16
+    # <note type="cf1">K30n1002_p0257a01-a23</note>
+    refs = []
+    e.xpath('./note').each { |n|
+      if n.key?('type') and n['type'].start_with? 'cf'
+        refs << n.content
+      end
+    }
+    if refs.empty?
+      ''
+    else
+      '修訂依據：' + refs.join('；') + '。'
+    end
+  end
+
+  def lem_note_rdg(lem)
+    r = ''
+    app = lem.parent
+    @pass << false
+    app.xpath('rdg').each { |rdg|
+      if rdg['wit'].include? @orig
+        s = traverse(rdg, 'back')
+        s = MISSING if s.empty?
+        r += @orig + s
+      end
+    }
+    @pass.pop
+    r += '。' unless r.empty?
+    r
+  end
+
+  def open_xml(fn)
+    s = File.read(fn)
+
+    if fn.include? 'T16n0657'
+      # 這個地方 雙行夾註 跨兩行偈頌
+      # 把 lb 移到 note 結束之前
+      # 讓 lg-row 先結束，再結束雙行夾註
+      s.sub!(/(<\/note>)(\n<lb n="0206b29" ed="T"\/>)/, '\2\1')
+    end
+
+    # <milestone unit="juan"> 前面的 lb 屬於新的這一卷
+    s.gsub!(/((?:<pb [^>]+>\n?)?<lb [^>]+>\n?)(<milestone [^>]*unit="juan"\/>)/, '\2\1')
+
+    doc = Nokogiri::XML(s)
+    doc.remove_namespaces!()
+    doc
+  end
+
+  def read_mod_notes(doc)
+    doc.xpath("//note[@type='mod']").each { |e|
+      @mod_notes << e['n']
+    }
+  end
+
+  def parse_xml(xml_fn)
+    @pass = [false]
+
+    doc = open_xml(xml_fn)
+    
+    e = doc.xpath("//titleStmt/title")[0]
+    @title = traverse(e, 'txt')
+    @title = @title.split()[-1]
+    
+    read_mod_notes(doc)
+
+    root = doc.root()
+    body = root.xpath("text/body")[0]
+    @pass = [true]
+
+    text = traverse(body)
+    text
+  end
+
+  def traverse(e, mode='html')
+    r = ''
+    e.children.each { |c| 
+      s = handle_node(c, mode)
+      r += s
+    }
+    r
+  end
+
+end