clooth · December 14, 2011 10:25
diff --git a/rubythehardway.rb b/rubythehardway.rb
 #
 # Learn Ruby The Hard Way
 # =======================
 # This tool will download you an offline
 # single-page html version of the book
 # =======================
 # NOTE: This was only does to test how well it performs.
 #       You should read the book on the site:
 #       http://ruby.learncodethehardway.org/book/
 #
 require 'curb'
 require 'nokogiri'

 # Methods for fetching pages
 def load_urls urls
  responses = {}
  m = Curl::Multi.new
  urls.each do |url|
    responses[url] = ""
    c = Curl::Easy.new(url) do|curl|
      curl.follow_location = true
      curl.on_body do |data|
        responses[url] << data
        data.size
      end
    end
    m.add(c)
  end
  m.perform

  responses
 end

 # Root url for the book
 root_url = "http://ruby.learncodethehardway.org/book/"

 # Chapter links
 chapter_urls = []

 # Fetch the main book index and get the links for each of the chapters
 main_page = load_urls([root_url])
 main_page = Nokogiri::HTML(main_page[root_url])
 main_page.css('.toctree-l1 a').each do |chapter_link|
  chapter_urls << root_url + chapter_link.attributes["href"].content
 end

 # Final page markup
 final_markup = <<FOOLOL
 <html>
 <head>
 <title>Learn Ruby The Hard Way</title>
 </head>
 <body>
 <h1>Learn Ruby The Hard Way Book</h1>
 <h2>Table of Contents</h2>
 <ul>%s</ul>
 %s
 </body>
 </html>
 FOOLOL

 # Fetch all the chapters
 chapter_pages = load_urls(chapter_urls)

 # Build table of contents
 table_of_contents = ""
 chapter_pages.each do |key, chapter_page|
  page  = Nokogiri::HTML(chapter_page)
  slug  = page.css('.section').first.attributes["id"].content
  title = page.css('.section h1').first.text.chop
  table_of_contents << "<li><a href=\"\##{slug}\">#{title}</a></li>\n"
 end

 # Build chapter contents
 chapter_contents = ""
 chapter_pages.each do |key, chapter_page|
  page    = Nokogiri::HTML(chapter_page)
  section = page.css('.section').first
  slug    = section.attributes["id"].content
  title   = section.css('h1').first.remove
  chapter_contents << "<a id=\"#{slug}\"></a><div class=\"section\"><h1>#{title.text}</h1><div>#{section.inner_html}</div></div>"
 end

 File.open('rubythehardway.html', 'w') do |file|
  file.write(final_markup % [table_of_contents, chapter_contents])
 end
	#
	# Learn Ruby The Hard Way
	# =======================
	# This tool will download you an offline
	# single-page html version of the book
	# =======================
	# NOTE: This was only does to test how well it performs.
	# You should read the book on the site:
	# http://ruby.learncodethehardway.org/book/
	#
	require 'curb'
	require 'nokogiri'

	# Methods for fetching pages
	def load_urls urls
	responses = {}
	m = Curl::Multi.new
	urls.each do \|url\|
	responses[url] = ""
	c = Curl::Easy.new(url) do\|curl\|
	curl.follow_location = true
	curl.on_body do \|data\|
	responses[url] << data
	data.size
	end
	end
	m.add(c)
	end
	m.perform

	responses
	end

	# Root url for the book
	root_url = "http://ruby.learncodethehardway.org/book/"

	# Chapter links
	chapter_urls = []

	# Fetch the main book index and get the links for each of the chapters
	main_page = load_urls([root_url])
	main_page = Nokogiri::HTML(main_page[root_url])
	main_page.css('.toctree-l1 a').each do \|chapter_link\|
	chapter_urls << root_url + chapter_link.attributes["href"].content
	end

	# Final page markup
	final_markup = <<FOOLOL
	<html>
	<head>
	<title>Learn Ruby The Hard Way</title>
	</head>
	<body>
	<h1>Learn Ruby The Hard Way Book</h1>
	<h2>Table of Contents</h2>
	<ul>%s</ul>
	%s
	</body>
	</html>
	FOOLOL

	# Fetch all the chapters
	chapter_pages = load_urls(chapter_urls)

	# Build table of contents
	table_of_contents = ""
	chapter_pages.each do \|key, chapter_page\|
	page = Nokogiri::HTML(chapter_page)
	slug = page.css('.section').first.attributes["id"].content
	title = page.css('.section h1').first.text.chop
	table_of_contents << "<li><a href=\"\##{slug}\">#{title}</a></li>\n"
	end

	# Build chapter contents
	chapter_contents = ""
	chapter_pages.each do \|key, chapter_page\|
	page = Nokogiri::HTML(chapter_page)
	section = page.css('.section').first
	slug = section.attributes["id"].content
	title = section.css('h1').first.remove
	chapter_contents << "<a id=\"#{slug}\"></a><div class=\"section\"><h1>#{title.text}</h1><div>#{section.inner_html}</div></div>"
	end

	File.open('rubythehardway.html', 'w') do \|file\|
	file.write(final_markup % [table_of_contents, chapter_contents])
	end