_Rakefile

desc "Check site for broken links"
task :check_links do
  BrokenLinkChecker.new('_site')
end


require 'pathname'

class BrokenLinkChecker

  def initialize(path)
    puts "Checking files in #{path} for broken links"
    @path = path
    @all_files = Dir.glob("#{@path}/**/*")
    html_files = Dir.glob("#{@path}/**/*.html")
    broken = html_files.inject({}) do |hash, file|
      broken_in_file = find_broken_links(file).uniq
      hash[file] = broken_in_file unless broken_in_file.empty?
      hash
    end
    puts "No broken links found" if broken.empty?
    broken.keys.each do |file|
      puts "#{file} has broken links #{broken[file].inspect}"
    end
  end

  protected

  def find_broken_links(file)
    page = File.read(file)
    links = page.scan(/href=(['"])(.+?)\1/).map(&:last)
    links.map do |link|
      if excluded?(link) || absolute_path?(link) || exists?(link, file)
        nil
      else
        link
      end
    end.compact
  end

  def excluded?(link)
    link =~ /^'\+(feeds|entry)\.link\+'$/
  end

  def absolute_path?(link)
    link =~ /^(https?|ftp|irc):/
  end

  def exists?(link, file)
    unless absolute_path?(link) || link =~ /^\//
      puts "WARNING: #{file} has a relative link to #{link}"
    end
    @all_files.include?("#{@path}#{link}") ||
      @all_files.include?("#{@path}#{link}index.html") ||
      @all_files.include?("#{@path}#{link}/index.html")
  end

end