RubyGems - aozora2html - Versions diffs - 0.7.1 → 0.9.0 - Mend

aozora2html 0.7.1 → 0.9.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (90) hide show

checksums.yaml +4 -4
data/bin/aozora2html +3 -50
data/lib/aozora2html.rb +0 -4
data/lib/aozora2html/accent_parser.rb +91 -0
data/lib/aozora2html/error.rb +16 -0
data/lib/aozora2html/header.rb +139 -0
data/lib/aozora2html/i18n.rb +17 -0
data/lib/aozora2html/ruby_buffer.rb +81 -0
data/lib/aozora2html/style_stack.rb +27 -0
data/lib/aozora2html/tag.rb +55 -0
data/lib/aozora2html/tag/accent.rb +39 -0
data/lib/aozora2html/tag/block.rb +19 -0
data/lib/aozora2html/tag/chitsuki.rb +15 -0
data/lib/aozora2html/tag/dakuten_katakana.rb +23 -0
data/lib/aozora2html/tag/decorate.rb +18 -0
data/lib/aozora2html/tag/dir.rb +15 -0
data/lib/aozora2html/tag/editor_note.rb +16 -0
data/lib/aozora2html/tag/embed_gaiji.rb +48 -0
data/lib/aozora2html/tag/font_size.rb +17 -0
data/lib/aozora2html/tag/gaiji.rb +11 -0
data/lib/aozora2html/tag/img.rb +21 -0
data/lib/aozora2html/tag/indent.rb +8 -0
data/lib/aozora2html/tag/inline.rb +13 -0
data/lib/aozora2html/tag/inline_caption.rb +15 -0
data/lib/aozora2html/tag/inline_font_size.rb +18 -0
data/lib/aozora2html/tag/inline_keigakomi.rb +15 -0
data/lib/aozora2html/tag/inline_yokogumi.rb +16 -0
data/lib/aozora2html/tag/jisage.rb +15 -0
data/lib/aozora2html/tag/jizume.rb +16 -0
data/lib/aozora2html/tag/kaeriten.rb +15 -0
data/lib/aozora2html/tag/keigakomi.rb +13 -0
data/lib/aozora2html/tag/kunten.rb +12 -0
data/lib/aozora2html/tag/midashi.rb +18 -0
data/lib/aozora2html/tag/multiline.rb +6 -0
data/lib/aozora2html/tag/multiline_caption.rb +15 -0
data/lib/aozora2html/tag/multiline_chitsuki.rb +8 -0
data/lib/aozora2html/tag/multiline_jisage.rb +8 -0
data/lib/aozora2html/tag/multiline_midashi.rb +22 -0
data/lib/aozora2html/tag/multiline_style.rb +17 -0
data/lib/aozora2html/tag/multiline_yokogumi.rb +17 -0
data/lib/aozora2html/tag/okurigana.rb +15 -0
data/lib/aozora2html/tag/oneline_chitsuki.rb +9 -0
data/lib/aozora2html/tag/oneline_indent.rb +2 -0
data/lib/aozora2html/tag/oneline_jisage.rb +7 -0
data/lib/aozora2html/tag/reference_mentioned.rb +46 -0
data/lib/aozora2html/tag/ruby.rb +98 -0
data/lib/aozora2html/tag/un_embed_gaiji.rb +24 -0
data/lib/aozora2html/tag_parser.rb +53 -0
data/lib/aozora2html/utils.rb +82 -0
data/lib/aozora2html/version.rb +1 -1
data/lib/aozora2html/yaml_loader.rb +31 -0
data/lib/extensions.rb +31 -0
data/lib/jstream.rb +68 -0
data/lib/t2hs.rb +485 -1490
data/test/test_aozora2html.rb +161 -57
data/test/test_aozora_accent_parser.rb +13 -2
data/test/test_command_parse.rb +213 -0
data/test/test_dakuten_katakana_tag.rb +5 -4
data/test/test_decorate_tag.rb +5 -4
data/test/test_dir_tag.rb +5 -4
data/test/test_editor_note_tag.rb +4 -4
data/test/test_exception.rb +4 -4
data/test/test_font_size_tag.rb +8 -7
data/test/test_gaiji_tag.rb +14 -7
data/test/test_header.rb +60 -0
data/test/test_img_tag.rb +5 -4
data/test/test_inline_caption_tag.rb +5 -4
data/test/test_inline_font_size_tag.rb +7 -6
data/test/test_inline_keigakomi_tag.rb +5 -4
data/test/test_inline_yokogumi_tag.rb +5 -4
data/test/test_jizume_tag.rb +7 -6
data/test/test_jstream.rb +5 -5
data/test/test_kaeriten_tag.rb +5 -4
data/test/test_keigakomi_tag.rb +8 -7
data/test/test_midashi_tag.rb +38 -0
data/test/test_multiline_caption_tag.rb +7 -6
data/test/test_multiline_midashi_tag.rb +12 -11
data/test/test_multiline_style_tag.rb +7 -6
data/test/test_multiline_yokogumi_tag.rb +7 -6
data/test/test_okurigana_tag.rb +5 -4
data/test/test_ruby_parse.rb +116 -0
data/test/test_ruby_tag.rb +5 -4
data/test/test_tag_parser.rb +10 -10
data/yml/accent_table.yml +240 -0
data/yml/command_table.yml +61 -0
data/yml/jis2ucs.yml +11234 -0
metadata +63 -6
data/lib/accent_tag.rb +0 -23
data/lib/aozora2html/jis2ucs.rb +0 -11237
data/lib/embed_gaiji_tag.rb +0 -34

checksums.yaml CHANGED

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 69f585eb774610409f4a56b1645bc2ebdf71ce35
-  data.tar.gz: c2c28ed11b58ca3d6df952f3b5709d14184e48f4
+  metadata.gz: ecb4e24d58fac283904c18a996d7f19d308f4d69
+  data.tar.gz: 0e3738df57189ff06736a508a6aae1c79a34667e
 SHA512:
-  metadata.gz: 9291546c4ff8e76d64e0e1eaa32474efcea5c6d952e8215841d2a0d1da0beccae37d01a4dd8d1689ebd9ff962b6511d1811e6e3b4d3b381be8d102ccd63356b4
-  data.tar.gz: 23235e3da67c52b05cfb648275888ee82239f94e1f47d3a78261188ae4eaa110e1ca18dc5d9b8af1113e054e00f56586976fb5fb7fb99628175c2f0c2625d4ce
+  metadata.gz: 84e3a6e4e4dc1d274238a122001995226e0712657a6111080d410dbc1ba1d33e80f1df646b4936bf9553e701e85e0fa9d50ab37ec0afab51dda1fbcdaf04af65
+  data.tar.gz: fb2f60e2bd5a4f6ee9d0e9c4597915431b4d69c58420a0bd894efee97f49061f317e240a8dd151c1e90591fd82de55b14cc5e4bb37221c2c7fed57f82d85ca6a

data/bin/aozora2html CHANGED

@@ -4,53 +4,6 @@ require 'aozora2html'
 require 'optparse'
 require "tempfile"
-# override Aozora2Html#push_chars
-#
-# Original Aozora2Html#push_chars does not convert "'" into '&#39;'; it's old behaivor
-# of CGI.escapeHTML().
-#
-class Aozora2Html
-  def push_chars(obj)
-    if obj.is_a?(Array)
-      obj.each{|x|
-        push_chars(x)
-      }
-    elsif obj.is_a?(String)
-      if obj.length == 1
-        obj = obj.gsub(/[&\"<>]/, {'&' => '&amp;', '"' => '&quot;', '<' => '&lt;', '>' => '&gt;'})
-      end
-      obj.each_char{|x|
-        push_char(x)
-      }
-    else
-      push_char(obj)
-    end
-  end
-  def dispatch_gaiji
-    hook = @stream.peek_char(0)
-    if hook ==  "［".encode("shift_jis")
-      read_char
-      # embed?
-      command,raw = read_to_nest("］".encode("shift_jis"))
-      try_emb = kuten2png(command)
-      if try_emb != command
-        try_emb
-      elsif command.match(/U\+([0-9A-F]{4,5})/) && Embed_Gaiji_tag.use_unicode
-        unicode_num = $1
-        ch = Embed_Gaiji_tag.new(self, nil, nil, command)
-        ch.unicode = unicode_num
-        ch
-      else
-        # Unemb
-        escape_gaiji(command)
-      end
-    else
-      "※".encode("shift_jis")
-    end
-  end
-end
 opt = OptionParser.new("Usage: aozora2html [options] <text file> [<html file>]\n")
 opt.on('--gaiji-dir DIR', 'setting gaiji directory')
 opt.on('--css-files FILES', 'setting css directory')
@@ -68,12 +21,12 @@ if options["css-files"]
 end
 if options["use-jisx0213"]
-  Embed_Gaiji_tag.use_jisx0213 = true
-  Accent_tag.use_jisx0213 = true
+  Aozora2Html::Tag::EmbedGaiji.use_jisx0213 = true
+  Aozora2Html::Tag::Accent.use_jisx0213 = true
 end
 if options["use-unicode"]
-  Embed_Gaiji_tag.use_unicode = true
+  Aozora2Html::Tag::EmbedGaiji.use_unicode = true
 end
 if ARGV.size < 1 || ARGV.size > 2

data/lib/aozora2html.rb CHANGED

@@ -1,9 +1,5 @@
 require "aozora2html/version"
-require "aozora2html/zip"
-require "aozora2html/jis2ucs"
 require 't2hs'
-require 'embed_gaiji_tag'
-require 'accent_tag'
 ## already defined in t2hs.rb
 class Aozora2Html

data/lib/aozora2html/accent_parser.rb ADDED

@@ -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(@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(@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 ADDED

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

data/lib/aozora2html/header.rb ADDED

@@ -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 |x|
+        code = x.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 ADDED

@@ -0,0 +1,17 @@
+# 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"
+    }
+    def self.t(msg, *args)
+      (MSG[msg].encode("shift_jis") % args)
+    end
+  end
+end

data/lib/aozora2html/ruby_buffer.rb ADDED

@@ -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(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 ADDED

@@ -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