aozora2html 0.7.0 → 2.0.0

data/lib/aozora2html/accent_parser.rb

@@ -0,0 +1,91 @@
+# encoding: utf-8
+require 'aozora2html/ruby_buffer'
+class Aozora2Html
+
+  # accent特殊文字を生かすための再帰呼び出し
+  class AccentParser < Aozora2Html
+
+    def initialize(input, endchar, chuuki, image)
+      if not(input.is_a?(Jstream))
+        raise ArgumentError, "tag_parser must supply Jstream as input"
+      end
+      @stream = input
+      @buffer = []
+      @ruby_buf = Aozora2Html::RubyBuffer.new
+      @chuuki_table = chuuki
+      @images = image # globalな環境を記録するアイテムは共有する必要あり
+      @endchar = endchar # 改行は越えられない <br />を出力していられない
+      @closed = nil # 改行での強制撤退チェックフラグ
+      @encount_accent = nil
+    end
+
+    def general_output # 出力は配列で返す
+      @ruby_buf.dump_into(@buffer)
+      if !@encount_accent
+        @buffer.unshift("〔".encode("shift_jis"))
+      end
+      if @closed and !@encount_accent
+        @buffer.push("〕".encode("shift_jis"))
+      elsif not(@closed)
+        @buffer.push("<br />\r\n")
+      end
+      @buffer
+    end
+
+    def parse
+      first = read_char
+      if found = Aozora2Html::ACCENT_TABLE[first]
+        if found2 = found[@stream.peek_char(0)]
+          if found2.is_a?(Hash)
+            if found3 = found2[@stream.peek_char(1)]
+              first = Aozora2Html::Tag::Accent.new(self, *found3)
+              @encount_accent = true
+              @chuuki_table[:accent] = true
+              read_char
+              read_char
+            end
+          elsif found2
+            first = Aozora2Html::Tag::Accent.new(self, *found2)
+            @encount_accent = true
+            read_char
+            @chuuki_table[:accent] = true
+          end
+        end
+      end
+      case first
+      when Aozora2Html::GAIJI_MARK
+        first = dispatch_gaiji
+      when "［".encode("shift_jis")
+        first = dispatch_aozora_command
+      when Aozora2Html::KU
+        assign_kunoji
+      when "《".encode("shift_jis")
+        first = apply_ruby
+      end
+      if first == "\r\n"
+        if @encount_accent
+          puts "警告(#{line_number}行目):アクセント分解の亀甲括弧の始めと終わりが、行中で揃っていません".encode("shift_jis")
+        end
+        throw :terminate
+      elsif first == "〕".encode("shift_jis")
+        @closed = true
+        throw :terminate
+      elsif first == RUBY_PREFIX
+        @ruby_buf.dump_into(@buffer)
+        @ruby_buf.protected = true
+      elsif first != "" and first != nil
+        illegal_char_check(first, line_number)
+        push_chars(first)
+      end
+    end
+
+    def process
+      catch(:terminate) do
+        loop do
+          parse
+        end
+      end
+      general_output
+    end
+  end
+end

data/lib/aozora2html/error.rb

@@ -0,0 +1,16 @@
+require "aozora2html/i18n"
+
+# 例外class
+class Aozora2Html
+  class Error < StandardError
+
+    def initialize(msg)
+      @message = msg
+    end
+
+    def message(line = 0)
+      I18n.t(:error_stop, line, @message)
+    end
+  end
+end
+

data/lib/aozora2html/header.rb

@@ -0,0 +1,139 @@
+# encoding: utf-8
+class Aozora2Html
+  class Header
+    def initialize()
+      @header = []
+    end
+
+    def push(line)
+      @header.push(line)
+    end
+
+    def out_header_info(hash, attr, true_name = nil)
+      found = hash[attr]
+      if found
+        "<h2 class=\"#{true_name or attr}\">#{found}</h2>\r\n"
+      else
+        ""
+      end
+    end
+
+    def header_element_type(string)
+      original = true
+      string.each_char do |ch|
+        code = ch.unpack("H*")[0]
+        if ("00" <= code and code <= "7f") or # 1byte
+          ("8140" <= code and code <= "8258") or # 1-1, 3-25
+          ("839f" <= code and code <= "8491") # 6-1, 7-81
+        # continue
+        else
+          original = false
+          break
+        end
+      end
+      if original
+        :original
+      elsif string.match(PAT_EDITOR)
+        :editor
+      elsif string.match(PAT_HENYAKU)
+        :henyaku
+      elsif string.match(PAT_TRANSLATOR)
+        :translator
+      end
+    end
+
+    def process_person(string, header_info)
+      type = header_element_type(string)
+      case type
+      when :editor
+        header_info[:editor] = string
+      when :translator
+        header_info[:translator] = string
+      when :henyaku
+        header_info[:henyaku] = string
+      else
+        type = :author
+        header_info[:author] = string
+      end
+      type
+    end
+
+    def build_title(header_info)
+      buf = [:author, :translator, :editor, :henyaku,
+             :title, :original_title,
+             :subtitle, :original_subtitle].map{|item| header_info[item]}.compact
+      buf_str = buf.join(" ")
+      "<title>#{buf_str}</title>"
+    end
+
+    def build_header_info
+      header_info = {:title => @header[0]}
+      case @header.length
+      when 2
+        process_person(@header[1], header_info)
+      when 3
+        if header_element_type(@header[1]) == :original
+          header_info[:original_title] = @header[1]
+          process_person(@header[2], header_info)
+        elsif process_person(@header[2], header_info) == :author
+          header_info[:subtitle] = @header[1]
+        else
+          header_info[:author] = @header[1]
+        end
+      when 4
+        if header_element_type(@header[1]) == :original
+          header_info[:original_title] = @header[1]
+        else
+          header_info[:subtitle] = @header[1]
+        end
+        if process_person(@header[3], header_info) == :author
+          header_info[:subtitle] = @header[2]
+        else
+          header_info[:author] = @header[2]
+        end
+      when 5
+        header_info[:original_title] = @header[1]
+        header_info[:subtitle] = @header[2]
+        header_info[:author] = @header[3]
+        if process_person(@header[4], header_info) == :author
+          raise Aozora2Html::Error, "parser encounted author twice"
+        end
+      when 6
+        header_info[:original_title] = @header[1]
+        header_info[:subtitle] = @header[2]
+        header_info[:original_subtitle] = @header[3]
+        header_info[:author] = @header[4]
+        if process_person(@header[5], header_info) == :author
+          raise Aozora2Html::Error, "parser encounted author twice"
+        end
+      end
+      header_info
+    end
+
+    def to_html
+      header_info = build_header_info()
+
+      # <title> 行を構築
+      html_title = build_title(header_info)
+
+      # 出力
+      out_buf = []
+      out_buf.push("<?xml version=\"1.0\" encoding=\"Shift_JIS\"?>\r\n<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML 1.1//EN\"\r\n    \"http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd\">\r\n<html xmlns=\"http://www.w3.org/1999/xhtml\" xml:lang=\"ja\" >\r\n<head>\r\n	<meta http-equiv=\"Content-Type\" content=\"text/html;charset=Shift_JIS\" />\r\n	<meta http-equiv=\"content-style-type\" content=\"text/css\" />\r\n")
+      $css_files.each do |css|
+        out_buf.push("\t<link rel=\"stylesheet\" type=\"text/css\" href=\"" + css + "\" />\r\n")
+      end
+      out_buf.push("\t#{html_title}\r\n	<script type=\"text/javascript\" src=\"../../jquery-1.4.2.min.js\"></script>\r\n  <link rel=\"Schema.DC\" href=\"http://purl.org/dc/elements/1.1/\" />\r\n	<meta name=\"DC.Title\" content=\"#{header_info[:title]}\" />\r\n	<meta name=\"DC.Creator\" content=\"#{header_info[:author]}\" />\r\n	<meta name=\"DC.Publisher\" content=\"#{AOZORABUNKO}\" />\r\n</head>\r\n<body>\r\n<div class=\"metadata\">\r\n")
+      out_buf.push("<h1 class=\"title\">#{header_info[:title]}</h1>\r\n" +
+                   out_header_info(header_info, :original_title) +
+                   out_header_info(header_info, :subtitle) +
+                   out_header_info(header_info, :original_subtitle) +
+                   out_header_info(header_info, :author) +
+                   out_header_info(header_info, :editor) +
+                   out_header_info(header_info, :translator) +
+                   out_header_info(header_info, :henyaku, "editor-translator"))
+      out_buf.push("<br />\r\n<br />\r\n</div>\r\n<div id=\"contents\" style=\"display:none\"></div><div class=\"main_text\">")
+      out_buf.join("")
+    end
+
+  end
+end

data/lib/aozora2html/i18n.rb

@@ -0,0 +1,28 @@
+# encoding: utf-8
+class Aozora2Html
+  class I18n
+    MSG = {
+      :tag_syntax_error => "注記を重ねる際の原則、「狭い範囲を先に、広い範囲を後に」が守られていません。リンク先の指針を参考に、書き方をあらためてください",
+      :undefined_header => "未定義な見出しです",
+      :use_crlf => "改行コードを、「CR+LF」にあらためてください",
+      :error_stop => "エラー(%d行目):%s. \r\n処理を停止します",
+      :invalid_font_size => "文字サイズの指定が不正です",
+      :unsupported_ruby  => "サポートされていない複雑なルビ付けです",
+      :warn_onebyte => "警告(%d行目):1バイトの「%s」が使われています",
+      :warn_chuki => "警告(%d行目):注記記号の誤用の可能性がある、「%s」が使われています",
+      :warn_jis_gaiji => "警告(%d行目):JIS外字「%s」が使われています",
+      :dont_crlf_in_style => "%s中に改行されました。改行をまたぐ要素にはブロック表記を用いてください",
+      :terminate_in_style => "%s中に本文が終了しました",
+      :invalid_closing => "%sを閉じようとしましたが、%s中ではありません",
+      :invalid_nesting => "%sを終了しようとしましたが、%s中です",
+      :dont_use_double_ruby => "同じ箇所に2つのルビはつけられません",
+      :dont_allow_triple_ruby => "1つの単語に3つのルビはつけられません",
+      :warn_unexpected_terminator => "警告(%d行目):予期せぬファイル終端",
+      :warn_undefined_command => "警告(%d行目):「%s」は未対応のコマンドのため無視します",
+    }
+
+    def self.t(msg, *args)
+      (MSG[msg].encode("shift_jis") % args)
+    end
+  end
+end

data/lib/aozora2html/ruby_buffer.rb

@@ -0,0 +1,81 @@
+class Aozora2Html
+  class RubyBuffer
+
+    # `｜`が来た時に真にする。ルビの親文字のガード用。
+    attr_accessor :protected
+
+    # @ruby_buf内の文字のchar_type
+    attr_accessor :char_type
+
+    def initialize(item=nil)
+      clear(item)
+    end
+
+    # バッファの初期化。引数itemがあるときはその1要素のバッファに、
+    # 引数がなければ`""`の1要素のバッファにする。
+    def clear(item=nil)
+      if item
+        @ruby_buf = [item]
+      else
+        @ruby_buf = [""]
+      end
+      @protected = nil
+      @char_type = nil
+    end
+
+    def empty?
+      @ruby_buf.empty?
+    end
+
+    def present?
+      !empty?
+    end
+
+    def to_a
+      @ruby_buf
+    end
+
+    def each(&block)
+      @ruby_buf.each(&block)
+    end
+
+    def last
+      @ruby_buf.last
+    end
+
+    def push(item)
+      @ruby_buf.push(item)
+    end
+
+    def length
+      @ruby_buf.length
+    end
+
+    def last_concat(item)
+      @ruby_buf.last.concat(item)
+    end
+
+    def last_is_string?
+      @ruby_buf.last.is_a?(String)
+    end
+
+    # buffer management
+    def dump_into(buffer)
+      if @protected
+        @ruby_buf.unshift(RUBY_PREFIX)
+        @protected = nil
+      end
+      top = @ruby_buf[0]
+      if top.is_a?(String) and buffer.last.is_a?(String)
+        buffer.last.concat(top)
+        buffer.concat(@ruby_buf[1,@ruby_buf.length])
+      else
+        buffer.concat(@ruby_buf)
+      end
+      clear
+      buffer
+    end
+
+  end
+end
+

data/lib/aozora2html/style_stack.rb

@@ -0,0 +1,27 @@
+class Aozora2Html
+  class StyleStack
+    def initialize
+      @stack = []
+    end
+
+    def push(elem)
+      @stack.push(elem)
+    end
+
+    def empty?
+      @stack.empty?
+    end
+
+    def pop
+      @stack.pop
+    end
+
+    def last
+      @stack.last
+    end
+
+    def last_command
+      @stack.last[0]
+    end
+  end
+end

data/lib/aozora2html/tag/accent.rb

@@ -0,0 +1,39 @@
+class Aozora2Html
+  class Tag
+    class Accent < Aozora2Html::Tag
+
+      @use_jisx0213 = nil
+
+      def self.use_jisx0213=(val)
+        @use_jisx0213 = val
+      end
+
+      def self.use_jisx0213
+        @use_jisx0213
+      end
+
+      include Aozora2Html::Tag::Inline
+      def initialize(parser, code, name)
+        @code = code
+        @name = name
+        super
+      end
+
+      def jisx0213_to_unicode(code)
+        Aozora2Html::JIS2UCS[code]
+      end
+
+      def char_type
+        :hankaku
+      end
+
+      def to_s
+        if Aozora2Html::Tag::Accent.use_jisx0213
+          jisx0213_to_unicode(@code.sub(%r|.*/|,"").to_sym)
+        else
+          "<img src=\"#{$gaiji_dir}#{@code}.png\" alt=\"" + GAIJI_MARK + "(#{@name})\" class=\"gaiji\" />"
+        end
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/block.rb

@@ -0,0 +1,19 @@
+class Aozora2Html
+  class Tag
+    module Block
+      def initialize(parser, *args)
+        if parser.block_allowed_context?
+          nil
+        else
+          syntax_error
+        end
+      end
+
+      # 必要に基づきmethod overrideする
+      def close_tag
+        "</div>"
+      end
+    end
+  end
+end
+

data/lib/aozora2html/tag/chitsuki.rb

@@ -0,0 +1,15 @@
+class Aozora2Html
+  class Tag
+    class Chitsuki < Aozora2Html::Tag::Indent
+
+      def initialize(parser, length)
+        @length = length
+        super
+      end
+
+      def to_s
+        '<div class="chitsuki_' + @length + '" style="text-align:right; margin-right: ' + @length + 'em">'
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/dakuten_katakana.rb

@@ -0,0 +1,23 @@
+# encoding: utf-8
+class Aozora2Html
+  class Tag
+    class DakutenKatakana < Aozora2Html::Tag
+      include Aozora2Html::Tag::Inline
+
+      def initialize(parser, n, katakana)
+        @n = n
+        @katakana = katakana
+        super
+      end
+
+      def char_type
+        :katakana
+      end
+
+      def to_s
+        "<img src=\"#{$gaiji_dir}/1-07/1-07-8#{@n}.png\" alt=\"" + "※(濁点付き片仮名「".encode("shift_jis") + @katakana + "」、1-07-8".encode("shift_jis") + "#{@n})\" class=\"gaiji\" />"
+      end
+    end
+  end
+end
+

data/lib/aozora2html/tag/decorate.rb

@@ -0,0 +1,18 @@
+class Aozora2Html
+  class Tag
+    class Decorate < Aozora2Html::Tag::ReferenceMentioned
+
+      def initialize(parser, target, html_class, html_tag)
+        @target = target
+        @close = "</#{html_tag}>"
+        @open = "<#{html_tag} class=\"#{html_class}\">"
+        super
+      end
+
+      def to_s
+        @open+@target.to_s+@close
+      end
+    end
+  end
+end
+

data/lib/aozora2html/tag/dir.rb

@@ -0,0 +1,15 @@
+class Aozora2Html
+  class Tag
+    class Dir < Aozora2Html::Tag::ReferenceMentioned
+
+      def initialize(parser, target)
+        @target = target
+        super
+      end
+
+      def to_s
+        "<span dir=\"ltr\">#{@target.to_s}</span>"
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/editor_note.rb

@@ -0,0 +1,16 @@
+# encoding: utf-8
+class Aozora2Html
+  class Tag
+    class EditorNote < Aozora2Html::Tag
+      include Aozora2Html::Tag::Inline
+      def initialize(parser, desc)
+        @desc = desc
+        super
+      end
+
+      def to_s
+        '<span class="notes">［＃'.encode("shift_jis") + @desc + '］</span>'.encode("shift_jis")
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/embed_gaiji.rb

@@ -0,0 +1,48 @@
+class Aozora2Html
+  class Tag
+    class EmbedGaiji < Aozora2Html::Tag::Gaiji
+      attr_accessor :unicode
+
+      @use_jisx0213 = nil
+      @use_unicode = nil
+
+      def self.use_jisx0213=(val)
+        @use_jisx0213 = val
+      end
+
+      def self.use_jisx0213
+        @use_jisx0213
+      end
+
+      def self.use_unicode=(val)
+        @use_unicode = val
+      end
+
+      def self.use_unicode
+        @use_unicode
+      end
+
+      def initialize(parser, folder, code, name, unicode_num = nil)
+        @folder = folder
+        @code = code
+        @name = name
+        @unicode = unicode_num
+        super
+      end
+
+      def jisx0213_to_unicode(code)
+        Aozora2Html::JIS2UCS[code]
+      end
+
+      def to_s
+        if Aozora2Html::Tag::EmbedGaiji.use_jisx0213 && @code
+          jisx0213_to_unicode(@code.to_sym)
+        elsif Aozora2Html::Tag::EmbedGaiji.use_unicode && @unicode
+          '&#x'+@unicode+';'
+        else
+          "<img src=\"#{$gaiji_dir}#{@folder}/#{@code}.png\" alt=\"" + GAIJI_MARK + "(#{@name})\" class=\"gaiji\" />"
+        end
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/font_size.rb

@@ -0,0 +1,17 @@
+class Aozora2Html
+  class Tag
+    class FontSize < Aozora2Html::Tag
+      include Aozora2Html::Tag::Block, Aozora2Html::Tag::Multiline
+
+      def initialize(parser, times, daisho)
+        @class = daisho.to_s + times.to_s
+        @style = Utils.create_font_size(times, daisho)
+        super
+      end
+
+      def to_s
+        "<div class=\"#{@class}\" style=\"font-size: #{@style};\">"
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/gaiji.rb

@@ -0,0 +1,11 @@
+class Aozora2Html
+  class Tag
+    class Gaiji < Aozora2Html::Tag
+      include Aozora2Html::Tag::Inline
+
+      def char_type
+        :kanji
+      end
+    end
+  end
+end

data/lib/aozora2html/tag/img.rb

@@ -0,0 +1,21 @@
+class Aozora2Html
+  class Tag
+    class Img < Aozora2Html::Tag
+      include Aozora2Html::Tag::Inline
+
+      def initialize(parser, filename, css_class, alt, width, height)
+        @filename = filename
+        @css_class = css_class
+        @alt = alt
+        @width = width
+        @height = height
+        super
+      end
+
+      def to_s
+        "<img class=\"#{@css_class}\" width=\"#{@width}\" height=\"#{@height}\" src=\"#{@filename}\" alt=\"#{@alt}\" />"
+      end
+    end
+  end
+end
+

data/lib/aozora2html/tag/indent.rb

@@ -0,0 +1,8 @@
+class Aozora2Html
+  class Tag
+    class Indent < Aozora2Html::Tag
+      include Aozora2Html::Tag::Block
+    end
+  end
+end
+

data/lib/aozora2html/tag/inline.rb

@@ -0,0 +1,13 @@
+# 全ての青空記法はHTML elementに変換される
+# したがって、block/inlineの区別がある
+# 全ての末端青空classはどちらかのmoduleをincludeする必要がある
+class Aozora2Html
+  class Tag
+    module Inline
+      def initialize(*args)
+        true
+      end
+    end
+  end
+end
+

data/lib/aozora2html/tag/inline_caption.rb

@@ -0,0 +1,15 @@
+class Aozora2Html
+  class Tag
+    class InlineCaption < Aozora2Html::Tag::ReferenceMentioned
+      def initialize(parser, target)
+        @target = target
+        super
+      end
+
+      def to_s
+        "<span class=\"caption\">#{@target.to_s}</span>"
+      end
+    end
+  end
+end
+