tdiary-hikidoc2markdown.rb

#!/usr/bin/env ruby
#
# Convert tDiary's documents from HikiDoc to Markdown
#
# Copyright (C) Tada, Tadashi <t@tdtds.jp>
# 
require 'open-uri'
require 'nokogiri'
require 'cgi'

def each_plugins
	while l = gets
		plugin = l.scan(/\[(.*)\]/).flatten.first
		next unless plugin
		if plugin =~ /^\[.*\]$/
			plugin.sub!(/^\[(.*)\]$/){$1} 
		end
		yield plugin
	end
end

def hiki_file(plugin); "#{plugin}.hiki"; end
def md_file(plugin); "#{plugin}.md"; end

def save_hiki(plugin)
	hiki = hiki_file(plugin)
	return if File.exist?(hiki)
	encoded_plugin = CGI.escape(plugin.encode('EUC-JP'))
	html = Nokogiri(open("http://docs.tdiary.org/ja/?c=edit;p=#{encoded_plugin}", &:read))
	open(hiki, 'w') do |f|
		f.write(html.css('div.textarea textarea').text)
	end
end

def convert_block(l)
	l.
	sub(/^!!! */, '### ').sub(/^!! */, '## ').sub(/^! */, '# ').
	sub(/^\*\*\* */, '    * ').sub(/^\*\* */, '  * ').sub(/^\* */, '* ').
	sub(/^:(.*?):(.*)/){"### #$1\n#$2"}.
	sub(/^<<<|^>>>/, '```').
	sub(/^""/, '> ').
	sub(/^ {1,3}([^ ]+)/){"    #$1"}
end

def convert_inline(l)
	l.gsub(/\[\[(.*?)\]\]/){
		label, url = $1.split(/\|/)
		url ? "[#{label}](#{url})" : "[[#{label}]]"
	}.
	gsub(/'''(.*?)'''/){"**#$1**"}.
	gsub(/<.*?>/){|t| l =~ /^ / ? t : "`#{t}`"}
end

def convert(plugin)
	open(md_file(plugin), 'w') do |w|
		open(hiki_file(plugin), &:read).each_line do |l|
			l = convert_block(l)
			l = convert_inline(l)
			w.write(l)
		end
	end
end

each_plugins do |plugin|
	save_hiki(plugin)
	convert(plugin)
end