aozora2html 0.7.1 → 3.0.0

data/lib/aozora2html/tag/ruby.rb

@@ -0,0 +1,202 @@
+# frozen_string_literal: true
+
+# complex ruby markup
+# if css3 is major supported, please fix ruby position with property "ruby-position"
+# see also: http://www.w3.org/TR/2001/WD-css3-ruby-20010216/
+class Aozora2Html
+  class Tag
+    # ルビ用
+    #
+    # 現状、under_rubyは無視しているのに注意
+    class Ruby < Aozora2Html::Tag::ReferenceMentioned
+      attr_accessor :ruby, :under_ruby
+      attr_reader :target
+
+      def initialize(parser, string, ruby, under_ruby = '')
+        @target = string
+        @ruby = ruby
+        @under_ruby = under_ruby
+        super
+      end
+
+      def to_s
+        "<ruby><rb>#{@target}</rb><rp>#{PAREN_BEGIN_MARK}</rp><rt>#{@ruby}</rt><rp>#{PAREN_END_MARK}</rp></ruby>"
+      end
+
+      # rubyタグの再割り当て
+      def self.rearrange_ruby(parser, targets, upper_ruby, under_ruby)
+        unless include_ruby?(targets)
+          return Aozora2Html::Tag::Ruby.new(parser, targets, upper_ruby, under_ruby)
+        end
+
+        new_targets = []
+        new_upper = if upper_ruby == ''
+                      []
+                    else
+                      upper_ruby
+                    end
+        new_under = if under_ruby == ''
+                      []
+                    else
+                      under_ruby
+                    end
+        if (new_upper.length > 1) && (new_under.length > 1)
+          raise Aozora2Html::Error, I18n.t(:dont_allow_triple_ruby)
+        end
+
+        targets.each do |x|
+          case x
+          when Aozora2Html::Tag::Ruby
+            raise Aozora2Html::Error, I18n.t(:dont_use_double_ruby) if x.target.is_a?(Array)
+
+            if x.ruby == ''
+              raise Aozora2Html::Error, I18n.t(:dont_use_double_ruby) unless new_under.is_a?(Array)
+
+              new_under.push(x.under_ruby)
+            else
+              raise Aozora2Html::Error, I18n.t(:dont_use_double_ruby) unless new_upper.is_a?(Array)
+
+              new_upper.push(x.ruby)
+            end
+            new_targets.push(x.target)
+          when Aozora2Html::Tag::ReferenceMentioned
+            if x.target.is_a?(Array)
+              # recursive
+              ruby2 = rearrange_ruby(parser, x.target, '', '')
+              target2, upper_ruby2, under_ruby2 = ruby2.target, ruby2.ruby, ruby2.under_ruby
+              # rotation!!
+              target2.each do |y|
+                tmp = x.dup
+                tmp.target = y
+                new_targets.push(tmp)
+              end
+              if new_under.is_a?(Array)
+                new_under.concat(under_ruby2)
+              elsif under_ruby2.to_s.length > 0
+                raise Aozora2Html::Error, I18n.t(:dont_use_double_ruby)
+              end
+              if new_upper.is_a?(Array)
+                new_upper.concat(upper_ruby2)
+              elsif upper_ruby2.to_s.length > 0
+                raise Aozora2Html::Error, I18n.t(:dont_use_double_ruby)
+              end
+            else
+              new_targets.push(x)
+              if new_under.is_a?(Array)
+                new_under.push('')
+              end
+              if new_upper.is_a?(Array)
+                new_upper.push('')
+              end
+            end
+          else
+            new_targets.push(x)
+            if new_under.is_a?(Array)
+              new_under.push('')
+            end
+            if new_upper.is_a?(Array)
+              new_upper.push('')
+            end
+          end
+        end
+
+        Aozora2Html::Tag::Ruby.new(parser, new_targets, new_upper, new_under)
+      end
+
+      # arrayがルビを含んでいればそのインデックスを返す
+      #
+      # @return [Integer, nil]
+      #
+      def self.include_ruby?(array)
+        array.index do |elt|
+          case elt
+          when Aozora2Html::Tag::Ruby
+            true
+          when Aozora2Html::Tag::ReferenceMentioned
+            if elt.target.is_a?(Array)
+              include_ruby?(elt.target)
+            else
+              elt.target.is_a?(Aozora2Html::Tag::Ruby)
+            end
+          end
+        end
+      end
+
+      # ----------------------------------------------------
+      #
+      # def gen_rt(string)
+      #   if string == ''
+      #     '<rt class="dummy_ruby"></rt>'
+      #   else
+      #     "<rt class=\"real_ruby\">#{string}</rt>"
+      #   end
+      # end
+      #
+      #
+      # complex ruby is waiting for IE support and CSS3 candidate
+      #   def to_s
+      #     ans = "<ruby class=\"complex_ruby\"><rbc>" # indicator of new version of aozora ruby
+      #     if @ruby.is_a?(Array) and @ruby.length > 0
+      #       # cell is used
+      #       @rbspan = @ruby.length
+      #     end
+      #     if @under_ruby.is_a?(Array) and @under_ruby.length > 0
+      #       # cell is used, but two way cell is not supported
+      #       if @rbspan
+      #         raise Aozora2Html::Error, I18n.t(:unsupported_ruby)
+      #       else
+      #         @rbspan = @under_ruby.length
+      #       end
+      #     end
+      #
+      #     # target
+      #     if @rbspan
+      #       @target.each{|x|
+      #         ans.concat("<rb>#{x.to_s}</rb>")
+      #       }
+      #     else
+      #       ans.concat("<rb>#{@target.to_s}</rb>")
+      #     end
+      #
+      #     ans.concat("</rbc><rtc>")
+      #
+      #     # upper ruby
+      #     if @ruby.is_a?(Array)
+      #       @ruby.each{|x|
+      #         ans.concat(gen_rt(x))
+      #       }
+      #     elsif @rbspan
+      #       if @ruby != ""
+      #         ans.concat("<rt class=\"real_ruby\" rbspan=\"#{@rbspan}\">#{@ruby}</rt>")
+      #       else
+      #         ans.concat("<rt class=\"dummy_ruby\" rbspan=\"#{@rbspan}\"></rt>")
+      #       end
+      #     else
+      #       ans.concat(gen_rt(@ruby))
+      #     end
+      #
+      #     ans.concat("</rtc>")
+      #
+      #     # under_ruby (if exists)
+      #     if @under_ruby.length > 0
+      #       ans.concat("<rtc>")
+      #       if @under_ruby.is_a?(Array)
+      #         @under_ruby.each{|x|
+      #           ans.concat(gen_rt(x))
+      #         }
+      #       elsif @rbspan
+      #         ans.concat("<rt class=\"real_ruby\" rbspan=\"#{@rbspan}\">#{@under_ruby}</rt>")
+      #       else
+      #         ans.concat(gen_rt(@under_ruby))
+      #       end
+      #       ans.concat("</rtc>")
+      #     end
+      #
+      #     # finalize
+      #     ans.concat("</ruby>")
+      #
+      #     ans
+      #   end
+    end
+  end
+end

data/lib/aozora2html/tag/un_embed_gaiji.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require_relative '../string_refinements'
+
+class Aozora2Html
+  class Tag
+    # 非埋め込み外字
+    class UnEmbedGaiji < Aozora2Html::Tag::Gaiji
+      def initialize(parser, desc)
+        @desc = desc
+        @escaped = false
+        super
+      end
+
+      using StringRefinements
+
+      def to_s
+        '<span class="notes">［'.to_sjis + @desc + '］</span>'.to_sjis
+      end
+
+      def escaped?
+        @escaped
+      end
+
+      def escape!
+        @escaped = true
+      end
+    end
+  end
+end

data/lib/aozora2html/tag.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+require_relative 'tag/inline'
+require_relative 'tag/block'
+require_relative 'tag/accent'
+require_relative 'tag/gaiji'
+require_relative 'tag/embed_gaiji'
+require_relative 'tag/un_embed_gaiji'
+require_relative 'tag/editor_note'
+require_relative 'tag/indent'
+require_relative 'tag/oneline_indent'
+require_relative 'tag/multiline'
+require_relative 'tag/multiline_style'
+require_relative 'tag/font_size'
+require_relative 'tag/jizume'
+require_relative 'tag/keigakomi'
+require_relative 'tag/multiline_yokogumi'
+require_relative 'tag/multiline_caption'
+require_relative 'tag/multiline_midashi'
+require_relative 'tag/jisage'
+require_relative 'tag/oneline_jisage'
+require_relative 'tag/multiline_jisage'
+require_relative 'tag/chitsuki'
+require_relative 'tag/oneline_chitsuki'
+require_relative 'tag/multiline_chitsuki'
+require_relative 'tag/reference_mentioned'
+require_relative 'tag/midashi'
+require_relative 'tag/ruby'
+require_relative 'tag/kunten'
+require_relative 'tag/kaeriten'
+require_relative 'tag/okurigana'
+require_relative 'tag/inline_keigakomi'
+require_relative 'tag/inline_yokogumi'
+require_relative 'tag/inline_caption'
+require_relative 'tag/inline_font_size'
+require_relative 'tag/decorate'
+require_relative 'tag/dakuten_katakana'
+require_relative 'tag/dir'
+require_relative 'tag/img'
+
+class Aozora2Html
+  # 変換される青空記法class
+  class Tag
+    # debug用
+    def inspect
+      to_s
+    end
+
+    def char_type
+      :else
+    end
+
+    def syntax_error
+      raise Aozora2Html::Error, I18n.t(:tag_syntax_error)
+    end
+  end
+end

data/lib/aozora2html/tag_parser.rb

@@ -0,0 +1,60 @@
+# frozen_string_literal: true
+
+require_relative 'ruby_buffer'
+
+class Aozora2Html
+  # 注記記法parser
+  #
+  # 青空記法の入れ子に対応（？）
+  class TagParser < Aozora2Html
+    def initialize(input, endchar, chuuki, image, gaiji_dir:) # rubocop:disable Lint/MissingSuper
+      unless input.is_a?(Jstream)
+        raise ArgumentError, 'tag_parser must supply Jstream as input'
+      end
+
+      @stream = input
+      @gaiji_dir = gaiji_dir
+      @buffer = TextBuffer.new
+      @ruby_buf = 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 # 改行を越えるべきか否か…
+      @section = :tail # 末尾処理と記法内はインデントをしないので等価
+      @raw = +'' # 外字変換前の生テキストを残したいことがあるらしい
+    end
+
+    # method override!
+    def read_char
+      c = @stream.read_char
+      @raw.concat(c)
+      c
+    end
+
+    def read_to_nest(endchar)
+      ans = super(endchar)
+      @raw.concat(ans[1])
+      ans
+    end
+
+    # 出力は[String,String]返しで！
+    def general_output
+      @ruby_buf.dump_into(@buffer)
+      ans = +''
+      @buffer.each do |s|
+        if s.is_a?(Aozora2Html::Tag::UnEmbedGaiji) && !s.escaped?
+          # 消してあった※を復活させて
+          ans.concat(GAIJI_MARK)
+        end
+        ans.concat(s.to_s)
+      end
+      [ans, @raw]
+    end
+
+    def process
+      catch(:terminate) do
+        parse
+      end
+      general_output
+    end
+  end
+end

data/lib/aozora2html/text_buffer.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+class Aozora2Html
+  # 本文テキスト用バッファ
+  #
+  # 要素はString以外も含む
+  # to_sすると文字列化できる
+  class TextBuffer < Array
+    # 行出力時に@bufferが空かどうか調べる
+    #
+    # @bufferの中身によって行末の出力が異なるため
+    #
+    # @return [true, false, :inline] 空文字ではない文字列が入っていればfalse、1行注記なら:inline、それ以外しか入っていなければtrue
+    #
+    def blank_type
+      each do |token|
+        return false if token.is_a?(String) && token != ''
+
+        if token.is_a?(Aozora2Html::Tag::OnelineIndent)
+          return :inline
+        end
+      end
+      true
+    end
+
+    # 行末で<br />を出力するべきかどうかの判別用
+    #
+    # @return [true, false] Multilineの注記しか入っていなければfalse、Multilineでも空文字でもない要素が含まれていればtrue
+    #
+    def terpri?
+      flag = true
+      each do |x|
+        case x
+        when Aozora2Html::Tag::Multiline
+          flag = false
+        when ''
+        # skip
+        else
+          return true
+        end
+      end
+
+      flag
+    end
+
+    def to_s
+      join
+    end
+  end
+end

data/lib/aozora2html/utils.rb

@@ -0,0 +1,156 @@
+# frozen_string_literal: true
+
+class Aozora2Html
+  # ユーティリティ関数モジュール
+  module Utils
+    using StringRefinements
+
+    KANJI_NUMS = '一二三四五六七八九〇'.to_sjis
+    KANJI_TEN = '十'.to_sjis
+    ZENKAKU_NUMS = '０-９'.to_sjis
+
+    def create_font_size(times, daisho)
+      size = case times
+             when 1
+               +''
+             when 2
+               +'x-'
+             else
+               raise Aozora2Html::Error, I18n.t(:invalid_font_size) unless times >= 3
+
+               +'xx-'
+             end
+
+      case daisho
+      when :dai
+        size << 'large'
+      when :sho
+        size << 'small'
+      else
+        raise Aozora2Html::Error, I18n.t(:invalid_font_size)
+      end
+
+      size
+    end
+    module_function :create_font_size
+
+    def create_midashi_tag(size)
+      if size.match(SIZE_SMALL)
+        'h5'
+      elsif size.match(SIZE_MIDDLE)
+        'h4'
+      elsif size.match(SIZE_LARGE)
+        'h3'
+      else
+        raise Aozora2Html::Error, I18n.t(:undefined_header)
+      end
+    end
+    module_function :create_midashi_tag
+
+    def create_midashi_class(type, tag)
+      normal_midashi_tag = {
+        'h5' => 'ko-midashi',
+        'h4' => 'naka-midashi',
+        'h3' => 'o-midashi'
+      }
+
+      dogyo_midashi_tag = {
+        'h5' => 'dogyo-ko-midashi',
+        'h4' => 'dogyo-naka-midashi',
+        'h3' => 'dogyo-o-midashi'
+      }
+
+      mado_midashi_tag = {
+        'h5' => 'mado-ko-midashi',
+        'h4' => 'mado-naka-midashi',
+        'h3' => 'mado-o-midashi'
+      }
+
+      case type
+      when :normal
+        normal_midashi_tag[tag]
+      when :dogyo
+        dogyo_midashi_tag[tag]
+      when :mado
+        mado_midashi_tag[tag]
+      else
+        raise Aozora2Html::Error, I18n.t(:undefined_header)
+      end
+    end
+    module_function :create_midashi_class
+
+    def convert_japanese_number(command)
+      tmp = command.tr(ZENKAKU_NUMS, '0-9')
+      tmp.tr!(KANJI_NUMS, '1234567890')
+      tmp.gsub!(/(\d)#{KANJI_TEN}(\d)/o) { "#{$1}#{$2}" }
+      tmp.gsub!(/(\d)#{KANJI_TEN}/o) { "#{$1}0" }
+      tmp.gsub!(/#{KANJI_TEN}(\d)/o) { "1#{$1}" }
+      tmp.gsub!(/#{KANJI_TEN}/o, '10')
+      tmp
+    end
+    module_function :convert_japanese_number
+
+    # 使うべきではない文字があるかチェックする
+    #
+    # 警告を出力するだけで結果には影響を与えない。警告する文字は以下:
+    #
+    # * 1バイト文字
+    # * `＃`ではなく`♯`
+    # * JIS(JIS X 0208)外字
+    #
+    # @return [void]
+    #
+    def illegal_char_check(char, line)
+      return unless char.is_a?(String)
+
+      code = char.unpack1('H*')
+      if (code == '21') ||
+         (code == '23') ||
+         ((code >= 'a1') && (code <= 'a5')) ||
+         ((code >= '28') && (code <= '29')) ||
+         (code == '5b') ||
+         (code == '5d') ||
+         (code == '3d') ||
+         (code == '3f') ||
+         (code == '2b') ||
+         ((code >= '7b') && (code <= '7d'))
+        puts I18n.t(:warn_onebyte, line, char)
+      end
+
+      if code == '81f2'
+        puts I18n.t(:warn_chuki, line, char)
+      end
+
+      if ((code >= '81ad') && (code <= '81b7')) ||
+         ((code >= '81c0') && (code <= '81c7')) ||
+         ((code >= '81cf') && (code <= '81d9')) ||
+         ((code >= '81e9') && (code <= '81ef')) ||
+         ((code >= '81f8') && (code <= '81fb')) ||
+         ((code >= '8240') && (code <= '824e')) ||
+         ((code >= '8259') && (code <= '825f')) ||
+         ((code >= '827a') && (code <= '8280')) ||
+         ((code >= '829b') && (code <= '829e')) ||
+         ((code >= '82f2') && (code <= '82fc')) ||
+         ((code >= '8397') && (code <= '839e')) ||
+         ((code >= '83b7') && (code <= '83be')) ||
+         ((code >= '83d7') && (code <= '83fc')) ||
+         ((code >= '8461') && (code <= '846f')) ||
+         ((code >= '8492') && (code <= '849e')) ||
+         ((code >= '84bf') && (code <= '84fc')) ||
+         ((code >= '8540') && (code <= '85fc')) ||
+         ((code >= '8640') && (code <= '86fc')) ||
+         ((code >= '8740') && (code <= '87fc')) ||
+         ((code >= '8840') && (code <= '889e')) ||
+         ((code >= '9873') && (code <= '989e')) ||
+         ((code >= 'eaa5') && (code <= 'eafc')) ||
+         ((code >= 'eb40') && (code <= 'ebfc')) ||
+         ((code >= 'ec40') && (code <= 'ecfc')) ||
+         ((code >= 'ed40') && (code <= 'edfc')) ||
+         ((code >= 'ee40') && (code <= 'eefc')) ||
+         ((code >= 'ef40') && (code <= 'effc'))
+        puts I18n.t(:warn_jis_gaiji, line, char)
+      end
+    end
+    module_function :illegal_char_check
+  end
+end

data/lib/aozora2html/version.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 class Aozora2Html
-  VERSION = "0.7.1"
+  VERSION = '3.0.0'
 end

data/lib/aozora2html/yaml_loader.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require 'yaml'
+
+class Aozora2Html
+  # YAML Loader class for Shift_JIS
+  class YamlLoader
+    def initialize(base_dir)
+      @base_dir = base_dir
+    end
+
+    def load(path)
+      tmp_data = YAML.load_file(File.join(@base_dir, path))
+      normalize_data(tmp_data)
+    end
+
+    using StringRefinements
+
+    def normalize_data(data)
+      case data
+      when String
+        data.to_sjis
+      when Hash
+        new_data = {}
+        data.each do |k, v|
+          new_data[normalize_data(k)] = normalize_data(v)
+        end
+        new_data
+      when Array
+        data.map { |item| normalize_data(item) }
+      else
+        # noop
+        data
+      end
+    end
+  end
+end

data/lib/aozora2html/zip.rb

@@ -1,5 +1,9 @@
+# frozen_string_literal: true
+
 require 'zip'
+
 class Aozora2Html
+  # zipファイルからテキストを抽出する
   class Zip
     def self.unzip(zipfilename, textfilename)
       ::Zip::File.open(zipfilename) do |zip_file|