osunyorg/admin

module Importers
  class Cleaner

    def self.html_to_string(html)
      h = html
      h = Importers::Cleaner.clean_html h
      h = ActionController::Base.helpers.strip_tags h
      h
    end

    def self.clean_string(string)
      string = string.to_s
      string = string.gsub(' ', ' ')
      string = string.gsub('&', '&')
      string = ActionView::Base.full_sanitizer.sanitize string
      string = remove_control_chars string
      string
    end
  
    def self.clean_html(html)
      # invalid byte sequence in UTF-8
      # https://stackoverflow.com/questions/32826781/invalid-byte-sequence-in-utf-8-when-sanitizing-wordpress-export-content
      html = html.force_encoding('UTF-8').scrub
      # Relaxed config : https://github.com/rgrove/sanitize/blob/main/lib/sanitize/config/relaxed.rb
      # iframe attributes from MDN : https://developer.mozilla.org/fr/docs/Web/HTML/Element/iframe
      fragment = Sanitize.fragment(html, Sanitize::Config.merge(Sanitize::Config::RELAXED,
        attributes: Sanitize::Config::RELAXED[:attributes].merge({
          all: Sanitize::Config::RELAXED[:attributes][:all].dup - ['class', 'style'],
          'a' => Sanitize::Config::RELAXED[:attributes]['a'].dup.delete('rel'),
          'iframe' => [
            'allow', 'allowfullscreen', 'allowpaymentrequest', 'csp', 'height', 'loading',
            'name', 'referrerpolicy', 'sandbox', 'src', 'srcdoc', 'width', 'align',
            'frameborder', 'longdesc', 'marginheight', 'marginwidth', 'scrolling'
          ]
        }),
        elements: Set.new(Sanitize::Config::RELAXED[:elements]) - ['div', 'style'] + ['iframe'],
        remove_contents: ['math', 'noembed', 'noframes', 'noscript', 'plaintext', 'script', 'style', 'svg', 'xmp'],
        whitespace_elements: {
          'div' => { :before => "", :after => "" }
        }
      ))
      fragment = Nokogiri::HTML5.fragment(fragment)
      if fragment.css('h1').any?
        # h1 => h2 ; h2 => h3 ; ...
        (1..5).to_a.reverse.each do |i|
          fragment.css("h#{i}").each { |element| element.name = "h#{i+1}" }
        end
      end
      html = fragment.to_html(preserve_newline: true)
      html = remove_control_chars html
      html
    end

    protected
  
    def self.remove_control_chars(string)
      # Control chars & LSEP are invisible or hard to detect
      string = string.delete("
", "
", "
", "’")
      string = string.gsub /\u2028/, ''
      string
    end

  end
end