skoji/densho_conv.rb

## densho_conv.rb
require 'uri'

class DenshoConv
  attr_accessor :file_prefix, :file_suffix, :title, :stylesheets, :count, :toc, :count_format, :replace_url, :outer_div_class

  def init_default_filters
    before_filter do
      |text|
      if @replace_url
        URI.extract(text, ['http', 'https']).each { |url|
          text.sub!(url, "<a href=\"#{url}\">#{url}</a>")
        }
      end
      text
    end

    head_inserter do
      ret = ""
      @stylesheets.each { |s|
       ret << "<link rel=\"stylesheet\" type=\"text/css\" href=\"#{s}\" />\n"
      }
      ret
    end
  end

  def initialize(&block)
    @count = 0
    @file_prefix=''
    @file_suffix=''
    @count_format="%03d"
    @outer_div_class='contents'
    @toc = {}
    init_default_filters
    yield self
    self
  end

  def current_filename
    "#{@file_prefix}#{@count_format%@count}#{@file_suffix}.html"
  end

  def finish_html
    if !@file.nil?
      print "</div>\n" if @outer_div_wrote
      print <<EOF
  </body>
</html>
EOF
      @file.close
      @file = nil
    end
  end

  def start_html
    print <<EOF
<?xml version="1.0" encoding="UTF-8"?>
<html xmlns="http://www.w3.org/1999/xhtml">
  <head>
	<title>#{@title}</title>
EOF
    head_inserters.each {
      |f|
      print f.call
    }

    print <<EOF
  </head>
  <body>
EOF
    if !@outer_div_class.nil?
      print "<div class=\"#{@outer_div_class}\">"
      @outer_div_wrote = true
    else
      @outer_div_wrote = nil
    end
  end


  def before_filters
    @before_filters ||= []
  end

  def after_filters
    @after_filters ||= []
  end

  def head_inserters
    @head_inserters ||= []
  end

  def before_filter (&block)
    before_filters << block
  end

  def after_filter (&block)
    after_filters << block
  end

  def head_inserter(&block)
    head_inserters << block
  end

  def switch_to_next_html(title = nil)
    @title = title if !title.nil?
    finish_html
    @toc[current_filename()] =  title unless title.nil?
    @file = File.open(current_filename, 'w+')
    start_html
    @count = @count + 1
  end

  def print (x)
    after_filters.each {
      |f|
      x = f.call(x)
    }
    @file.print x
  end

  def send_if_command(l)
      l.strip =~ /^(\w*?):(.*?)$/
      if (!$1.nil? && respond_to?($1))
        send($1.strip,"#{$2}")
        true
      else
        false
      end
  end

  def parse_and_write(l)
      if !send_if_command(l) && l.strip.length > 0
        print l.strip
        print "<br />" unless l.strip =~  /(<\/[\w]+?>$|<[\w\s'"=]+?\/>$)/
        print "\n"
      end
  end

  def run
    while l = gets
      before_filters.each {
        |f|
        l = f.call(l)
      }
      parse_and_write l
    end
    finish_html
    self


  end
end

## sample.rb
# -*- coding: utf-8 -*-
require 'pp'
load 'densho_conv.rb'

Encoding.default_external='utf-8'

pp DenshoConv.new {
  |t|
  t.file_suffix = '_file'
  t.count_format='%02d'
  t.stylesheets = ['style.css']
  t.title = 'test'
  t.replace_url = true

  class << t
    def part(x)
      switch_to_next_html x.strip
      print "<h1>#{x.strip}</h1>\n"
    end

    def section(x)
      print "<h2>#{x.strip}</h2>\n"
    end

    def speak(x)
      x =~ /^(.*?)\s(.*)$/
      print "<span class='name'>#{$1.strip}</span><div class='comment'>#{$2.strip}</div>\n"
    end
  end
}.run.toc


## sample.txt
part: はじめに
section: このファイルの概要

<p>これはテスト用のサンプルです
頭にコマンドがなければそのままHTML内にでてきます</p>

part: ここから次のファイル
section: 本題

<p>ここで、次のファイルがはじまります。</p>

speak: 発言者 発言
speak: 発言者2 発言

<p> このサンプルでのspeakは、こういう対談形式を手っ取り早くマークアップするのに使えます</p>
<p> @replace_urlをtrueにすれば、https://gist.github.com/gists/1015999/edit ←こういう文字列を&gt;a&lt;で囲ってhtmlにします。</p>
	require 'uri'

	class DenshoConv
	attr_accessor :file_prefix, :file_suffix, :title, :stylesheets, :count, :toc, :count_format, :replace_url, :outer_div_class

	def init_default_filters
	before_filter do
	\|text\|
	if @replace_url
	URI.extract(text, ['http', 'https']).each { \|url\|
	text.sub!(url, "<a href=\"#{url}\">#{url}</a>")
	}
	end
	text
	end

	head_inserter do
	ret = ""
	@stylesheets.each { \|s\|
	ret << "<link rel=\"stylesheet\" type=\"text/css\" href=\"#{s}\" />\n"
	}
	ret
	end
	end

	def initialize(&block)
	@count = 0
	@file_prefix=''
	@file_suffix=''
	@count_format="%03d"
	@outer_div_class='contents'
	@toc = {}
	init_default_filters
	yield self
	self
	end

	def current_filename
	"#{@file_prefix}#{@count_format%@count}#{@file_suffix}.html"
	end

	def finish_html
	if !@file.nil?
	print "</div>\n" if @outer_div_wrote
	print <<EOF
	</body>
	</html>
	EOF
	@file.close
	@file = nil
	end
	end

	def start_html
	print <<EOF
	<?xml version="1.0" encoding="UTF-8"?>
	<html xmlns="http://www.w3.org/1999/xhtml">
	<head>
	<title>#{@title}</title>
	EOF
	head_inserters.each {
	\|f\|
	print f.call
	}

	print <<EOF
	</head>
	<body>
	EOF
	if !@outer_div_class.nil?
	print "<div class=\"#{@outer_div_class}\">"
	@outer_div_wrote = true
	else
	@outer_div_wrote = nil
	end
	end


	def before_filters
	@before_filters \|\|= []
	end

	def after_filters
	@after_filters \|\|= []
	end

	def head_inserters
	@head_inserters \|\|= []
	end

	def before_filter (&block)
	before_filters << block
	end

	def after_filter (&block)
	after_filters << block
	end

	def head_inserter(&block)
	head_inserters << block
	end

	def switch_to_next_html(title = nil)
	@title = title if !title.nil?
	finish_html
	@toc[current_filename()] = title unless title.nil?
	@file = File.open(current_filename, 'w+')
	start_html
	@count = @count + 1
	end

	def print (x)
	after_filters.each {
	\|f\|
	x = f.call(x)
	}
	@file.print x
	end

	def send_if_command(l)
	l.strip =~ /^(\w?):(.?)$/
	if (!$1.nil? && respond_to?($1))
	send($1.strip,"#{$2}")
	true
	else
	false
	end
	end

	def parse_and_write(l)
	if !send_if_command(l) && l.strip.length > 0
	print l.strip
	print "<br />" unless l.strip =~ /(<\/[\w]+?>$\|<[\w\s'"=]+?\/>$)/
	print "\n"
	end
	end

	def run
	while l = gets
	before_filters.each {
	\|f\|
	l = f.call(l)
	}
	parse_and_write l
	end
	finish_html
	self


	end
	end
	# -- coding: utf-8 --
	require 'pp'
	load 'densho_conv.rb'

	Encoding.default_external='utf-8'

	pp DenshoConv.new {
	\|t\|
	t.file_suffix = '_file'
	t.count_format='%02d'
	t.stylesheets = ['style.css']
	t.title = 'test'
	t.replace_url = true

	class << t
	def part(x)
	switch_to_next_html x.strip
	print "<h1>#{x.strip}</h1>\n"
	end

	def section(x)
	print "<h2>#{x.strip}</h2>\n"
	end

	def speak(x)
	x =~ /^(.?)\s(.)$/
	print "<span class='name'>#{$1.strip}</span><div class='comment'>#{$2.strip}</div>\n"
	end
	end
	}.run.toc
	part: はじめに
	section: このファイルの概要

	<p>これはテスト用のサンプルです
	頭にコマンドがなければそのままHTML内にでてきます</p>

	part: ここから次のファイル
	section: 本題

	<p>ここで、次のファイルがはじまります。</p>

	speak: 発言者発言
	speak: 発言者2 発言

	<p> このサンプルでのspeakは、こういう対談形式を手っ取り早くマークアップするのに使えます</p>
	<p> @replace_urlをtrueにすれば、https://gist.github.com/gists/1015999/edit ←こういう文字列を>a<で囲ってhtmlにします。</p>