Conversión de algunos tags problemáticos para el Editor Grammata en ficheros epubs con Loofah

style2tag_scrubber.rb

require "loofah"

style2tag = Loofah::Scrubber.new do |node|
  if node.attributes and node.attr('style')
    if node.attr('style')['font-style:italic'] and
        node.attr('style')['font-weight:bold']
      new_node = node.dup
      new_node.name = 'i'
      new_node.remove_attribute('style')

      node.children = new_node
      node.name = 'b'
      node.remove_attribute('style')
    elsif node.attr('style')['font-style:italic']
      node.name = 'i'
      node.remove_attribute('style')
    elsif node.attr('style')['font-weight:bold']
      node.name = 'b'
      node.remove_attribute('style')
    end
  end
end

Dir.foreach('in') do |entry|
  if File.file?("in/#{entry}")
    html = nil
    File.open("in/#{entry}", "r") do |f|
      html = f.read
    end

    html_sanitized = Loofah.document(html).scrub!(style2tag)

    File.open("out/#{entry}", "w") do |f|
      f.write(html_sanitized)
    end
  end
end