mizuho 0.9.6

data/bin/mizuho

@@ -0,0 +1,47 @@
+#!/usr/bin/env ruby
+require File.expand_path(File.dirname(__FILE__) + "/../lib/mizuho")
+require 'optparse'
+require 'mizuho/generator'
+
+$KCODE = 'UTF-8'
+options = {}
+parser = OptionParser.new do |opts|
+	opts.banner = "Usage: mizuho [options] INPUT"
+	opts.separator ""
+	
+	opts.separator "Options:"
+	opts.on("-t", "--template FILE", String, "Specify a template file to use.") do |value|
+		options[:template] = value
+	end
+	opts.on("-m", "--multi-page", "Generate one file per chapter.") do |value|
+		options[:multi_page] = value
+	end
+	opts.on("--icons-dir DIR", "Specify the directory in which icons\n" <<
+		"#{' ' * 37}should be searched. Defaults to\n" <<
+		"#{' ' * 37}'images/icons'.") do |value|
+		options[:icons_dir] = value
+	end
+	opts.on("-o", "--output FILE", String, "Specify the output filename.") do |value|
+		options[:output] = value
+	end
+end
+begin
+	parser.parse!
+rescue OptionParser::ParseError => e
+	puts e
+	puts
+	puts "Please see '--help' for valid options."
+	exit 1
+end
+
+begin
+	if ARGV.empty?
+		puts parser
+		exit 1
+	else
+		Mizuho::Generator.new(ARGV[0], options).start
+	end
+rescue Mizuho::GenerationError
+	STDERR.puts "*** ERROR"
+	exit 2
+end

data/bin/mizuho-asciidoc

@@ -0,0 +1,4 @@
+#!/usr/bin/env ruby
+require File.expand_path(File.dirname(__FILE__) + "/../lib/mizuho")
+require 'mizuho/source_highlight'
+exec("python", Mizuho::ASCIIDOC, *ARGV)

data/lib/mizuho.rb

@@ -0,0 +1,10 @@
+module Mizuho
+	SOURCE_ROOT = File.expand_path(File.dirname(__FILE__) + "/..")
+	LIBDIR      = "#{SOURCE_ROOT}/lib"
+	ASCIIDOC    = "#{SOURCE_ROOT}/asciidoc/asciidoc.py"
+	
+	if $LOAD_PATH.first != LIBDIR
+		$LOAD_PATH.unshift(LIBDIR)
+		$LOAD_PATH.uniq!
+	end
+end if !defined?(Mizuho::SOURCE_ROOT)

data/lib/mizuho/chapter.rb

@@ -0,0 +1,54 @@
+module Mizuho
+
+class Chapter
+	# This may be nil, indicating that this chapter is the preamble.
+	attr_accessor :heading
+	attr_accessor :filename
+	attr_accessor :contents
+	
+	def is_preamble?
+		return heading.nil?
+	end
+	
+	def title
+		if @heading
+			return @heading.title
+		else
+			return nil
+		end
+	end
+	
+	def title_without_numbers
+		if @heading
+			return @heading.title_without_numbers
+		else
+			return nil
+		end
+	end
+	
+	def plain_title
+		if @heading
+			return @heading.plain_title
+		else
+			return nil
+		end
+	end
+	
+	def basename
+		return File.basename(filename)
+	end
+	
+	def anchor
+		if @heading
+			return @heading.anchor
+		else
+			return nil
+		end
+	end
+	
+	def anchor_id
+		return anchor.sub(/^#/, '')
+	end
+end
+
+end

data/lib/mizuho/generator.rb

@@ -0,0 +1,134 @@
+require 'optparse'
+require 'digest/sha1'
+require 'mizuho'
+require 'mizuho/parser'
+require 'mizuho/template'
+require 'mizuho/source_highlight'
+
+module Mizuho
+
+class GenerationError < StandardError
+end
+
+class Generator
+	def initialize(input, options = {})
+		@input_file = input
+		@output_name = options[:output]
+		@template = locate_template_file(options[:template])
+		@multi_page = options[:multi_page]
+		@icons_dir = options[:icons_dir]
+		@conf_file = options[:conf_file]
+	end
+	
+	def start
+		output_filename = determine_output_filename(@input_file, @output_name)
+		self.class.run_asciidoc(@input_file, output_filename, @icons_dir, @conf_file)
+		if @template
+			apply_template(output_filename, @input_file, @output_name, @template, @multi_page)
+		end
+	end
+	
+	def self.run_asciidoc(input, output, icons_dir = nil, conf_file = nil)
+		args = ["python", ASCIIDOC, "-a", "toc", "-a", "icons"]
+		if icons_dir
+			args << "-a"
+			args << "iconsdir=#{icons_dir}"
+		end
+		if conf_file
+			# With the splat operator we support a string and an array of strings.
+			[*conf_file].each do |cf|
+				args << "-f"
+				args << cf
+			end
+		end
+		args += ["-n", "-o", output, input]
+		if !system(*args)
+			raise GenerationError, "Asciidoc failed."
+		end
+	end
+
+private
+	def locate_template_file(template_name)
+		if template_name.nil?
+			return "#{ROOT}/templates/asciidoc.html.erb"
+		elsif template_name =~ %r{[/.]}
+			# Looks like a filename.
+			return template_name
+		else
+			return "#{ROOT}/templates/#{template_name}.html.erb"
+		end
+	end
+	
+	def determine_output_filename(input, output = nil, chapter_id = nil)
+		if chapter_id
+			if output
+				dirname = File.dirname(output)
+				extname = File.extname(output)
+				basename = File.basename(output, extname)
+				filename = File.join(dirname, "#{basename}-#{chapter_id}#{extname}")
+			else
+				dirname = File.dirname(input)
+				basename = File.basename(input, File.extname(input))
+				filename = File.join(dirname, "#{basename}-#{chapter_id}.html")
+			end
+		else
+			if output
+				filename = output
+			else
+				dirname = File.dirname(input)
+				basename = File.basename(input, File.extname(input))
+				filename = File.join(dirname, "#{basename}.html")
+			end
+		end
+		return File.expand_path(filename)
+	end
+	
+	def apply_template(asciidoc_file, input_file, output_name, template_file, multi_page)
+		parser = Parser.new(asciidoc_file)
+		if multi_page
+			File.unlink(asciidoc_file)
+			assign_chapter_filenames_and_heading_basenames(parser.chapters, input_file, output_name)
+			parser.chapters.each_with_index do |chapter, i|
+				template = Template.new(template_file,
+					:multi_page? => true,
+					:title => parser.title,
+					:table_of_contents => parser.table_of_contents,
+					:contents => chapter.contents,
+					:is_preamble? => chapter.heading.nil?,
+					:chapters => parser.chapters,
+					:prev_chapter => (i <= 1) ? nil : parser.chapters[i - 1],
+					:current_chapter => chapter,
+					:next_chapter => parser.chapters[i + 1])
+				template.save(chapter.filename)
+			end
+		else
+			template = Template.new(template_file,
+				:multi_page? => false,
+				:title => parser.title,
+				:table_of_contents => parser.table_of_contents,
+				:contents => parser.contents)
+			template.save(asciidoc_file)
+		end
+	rescue Template::Error => e
+		STDERR.puts("*** #{template_file}:\n#{e}")
+		exit 1
+	end
+	
+	def assign_chapter_filenames_and_heading_basenames(chapters, input_file, output_name)
+		chapters.each_with_index do |chapter, i|
+			if chapter.is_preamble?
+				chapter.filename = determine_output_filename(input_file, output_name)
+			else
+				title_sha1 = Digest::SHA1.hexdigest(chapter.title_without_numbers)
+				chapter.filename = determine_output_filename(input_file,
+					output_name, title_sha1.slice(0..7))
+				chapter.heading.basename = File.basename(chapter.filename)
+				chapter.heading.each_descendant do |h|
+					h.basename = File.basename(chapter.filename)
+				end
+			end
+		end
+	end
+end
+
+end

data/lib/mizuho/heading.rb

@@ -0,0 +1,46 @@
+require 'cgi'
+
+module Mizuho
+
+class Heading
+	# The heading title, as HTML. Non-HTML special characters are already escaped.
+	attr_accessor :title
+	attr_accessor :level
+	attr_accessor :basename
+	attr_accessor :anchor
+	attr_accessor :parent
+	attr_accessor :children
+	
+	def initialize
+		@children = []
+	end
+	
+	def find_parent_with_level(level)
+		h = self
+		while h && h.level != level
+			h = h.parent
+		end
+		return h
+	end
+	
+	# The heading title without section number, as HTML. Non-HTML special characters
+	# are already escaped.
+	def title_without_numbers
+		return title.sub(/^(\d+\.)+ /, '')
+	end
+	
+	# The heading title without section number, as plain text. Contains no HTML
+	# elements. Non-HTML special characters are NOT escaped.
+	def plain_title
+		return CGI::unescapeHTML(title_without_numbers.gsub(%r{</?[a-z]+>}i, ''))
+	end
+	
+	def each_descendant(&block)
+		children.each do |h|
+			block.call(h)
+			h.each_descendant(&block)
+		end
+	end
+end
+
+end

data/lib/mizuho/parser.rb

@@ -0,0 +1,103 @@
+require 'cgi'
+require 'mizuho/heading'
+require 'mizuho/chapter'
+
+module Mizuho
+
+# This class can parse the raw Asciidoc XHTML output, and extract the title, raw
+# contents (without layout) and other information from it.
+class Parser
+	attr_reader :filename
+	
+	# The document's title.
+	attr_reader :title
+	# The document's table of contents, represented in a tree structure
+	# by Heading objects.
+	attr_reader :table_of_contents
+	# The document's raw contents, without any layout.
+	attr_reader :contents
+
+	# Parse the given file.
+	def initialize(filename)
+		@filename = filename
+		
+		@contents = File.read(filename)
+		
+		# Extract the title.
+		@contents =~ %r{<title>(.*?)</title>}
+		@title = CGI::unescapeHTML($1)
+		
+		# Get rid of the Asciidoc layout and unwanted elements.
+		if !@contents.sub!(/\A.*?(<div id="preamble">)/m, '\1')
+			# There's no preamble, so strip everything until the
+			# end of the TOC div.
+			@contents.sub!(%r(\A.*?</noscript>[\r\n\s]*</div>[\r\n\s]*</div>)m, '')
+		end
+		@contents.sub!(/<div id="footer">.*/m, '')
+		@contents.gsub!(%r{<div style="clear:left"></div>}, '')
+		
+		# Extract table of contents.
+		parse_table_of_contents(@contents)
+	end
+	
+	# Returns the individual chapters as an array of Chapter objects. The
+	# first Chapter object represents the preamble.
+	def chapters
+		@chapters ||= parse_chapters(@contents)
+	end
+
+private
+	def parse_table_of_contents(html)
+		@headings = []
+		current_heading = toplevel_heading = Heading.new
+		current_heading.title = @title
+		current_heading.level = 1
+		offset = 0
+		while true
+			offset = html.index(%r{<h(\d) id="(.*?)">(.*?)</h\d>}, offset)
+			break if offset.nil?
+			offset = $~.end(0)
+			level  = $1.to_i
+			anchor = CGI::unescapeHTML($2)
+			title  = $3
+			
+			new_heading = Heading.new
+			new_heading.title = title
+			new_heading.level = level
+			new_heading.anchor = "##{anchor}"
+			new_heading.parent = current_heading.find_parent_with_level(level - 1)
+			new_heading.parent.children << new_heading
+			current_heading = new_heading
+			@headings << new_heading
+		end
+		@table_of_contents = toplevel_heading.children
+	end
+	
+	def parse_chapters(html)
+		if !defined?(Hpricot)
+			require 'rubygems'
+			require 'hpricot'
+		end
+		doc = Hpricot(contents)
+		result = []
+		
+		has_preamble = !(doc / '#preamble').empty?
+		
+		# TODO: fix cross-references
+		(doc / 'div.sectionbody').each_with_index do |elem, i|
+			chapter = Chapter.new
+			if has_preamble
+				if i > 0
+					chapter.heading = @table_of_contents[i - 1]
+				end
+			else
+				chapter.heading = @table_of_contents[i]
+			end
+			chapter.contents = elem.inner_html
+			result << chapter
+		end
+		return result
+	end
+end
+
+end

data/lib/mizuho/source_highlight.rb

@@ -0,0 +1,2 @@
+require 'mizuho'
+ENV['PATH'] += ":" + File.join(Mizuho::SOURCE_ROOT, "source-highlight")

data/lib/mizuho/template.rb

@@ -0,0 +1,50 @@
+require 'erb'
+require 'mizuho/parser'
+
+module Mizuho
+
+class Template
+	class Error < StandardError
+	end
+	
+	def initialize(template_file, options)
+		@template_file = template_file
+		@base_dir = File.expand_path(File.dirname(@template_file))
+		@options = options
+		options.each_key do |key|
+			raise "All option keys must be symbols." if !key.is_a?(Symbol)
+			raise "Invalid key name '#{key}'." if key.to_s !~ /\A[a-z0-9_]+\??\Z/i
+			eval %{
+				def #{key}
+					@options[:#{key}]
+				end
+			}
+		end
+		apply
+	end
+	
+	def save(output_file)
+		File.open(output_file, 'w') do |f|
+			f.write(@output_contents)
+		end
+	end
+
+protected
+	include ERB::Util
+	
+	def include_file(basename)
+		return File.read(File.join(@base_dir, basename))
+	end
+	
+private
+	def get_binding
+		return binding
+	end
+	
+	def apply
+		erb = ERB.new(File.read(@template_file), nil, '-')
+		@output_contents = eval(erb.src, get_binding, @template_file)
+	end
+end
+
+end