Contents

# -*- coding: utf-8; frozen_string_literal: true -*-
#
#--
# Copyright (C) 2009-2019 Thomas Leitner <[e-mail]>
#
# This file is part of kramdown which is licensed under the MIT.
#++
#

require 'kramdown/converter'

module Kramdown

  module Converter

    # Removes all block (and optionally span) level HTML tags from the element tree.
    #
    # This converter can be used on parsed HTML documents to get an element tree that will only
    # contain native kramdown elements.
    #
    # *Note* that the returned element tree may not be fully conformant (i.e. the content models of
    # *some elements may be violated)!
    #
    # This converter modifies the given tree in-place and returns it.
    class RemoveHtmlTags < Base

      def initialize(root, options)
        super
        @options[:template] = ''
      end

      def convert(el)
        real_el, el = el, el.value if el.type == :footnote

        children = el.children.dup
        index = 0
        while index < children.length
          if children[index].type == :xml_pi ||
              (children[index].type == :html_element && (children[index].value == 'style' ||
                                                         children[index].value == 'script'))
            children[index..index] = []
          elsif children[index].type == :html_element &&
              ((@options[:remove_block_html_tags] && children[index].options[:category] == :block) ||
               (@options[:remove_span_html_tags] && children[index].options[:category] == :span))
            children[index..index] = children[index].children
          else
            convert(children[index])
            index += 1
          end
        end
        el.children = children
        real_el || el
      end

    end

  end
end

Version data entries

2 entries across 2 versions & 1 rubygems

Version	Path
kramdown-2.5.1	lib/kramdown/converter/remove_html_tags.rb
kramdown-2.5.0	lib/kramdown/converter/remove_html_tags.rb