app/lib/toc_generator.rb

   1 # frozen_string_literal: true
   2
   3 class TOCGenerator
   4   TARGET_ELEMENTS = %w(h1 h2 h3 h4 h5 h6).freeze
   5   LISTED_ELEMENTS = %w(h2 h3).freeze
   6
   7   class Section
   8     attr_accessor :depth, :title, :children, :anchor
   9
  10     def initialize(depth, title, anchor)
  11       @depth    = depth
  12       @title    = title
  13       @children = []
  14       @anchor   = anchor
  15     end
  16
  17     delegate :<<, to: :children
  18   end
  19
  20   def initialize(source_html)
  21     @source_html = source_html
  22     @processed   = false
  23     @target_html = ''
  24     @headers     = []
  25     @slugs       = Hash.new { |h, k| h[k] = 0 }
  26   end
  27
  28   def html
  29     parse_and_transform unless @processed
  30     @target_html
  31   end
  32
  33   def toc
  34     parse_and_transform unless @processed
  35     @headers
  36   end
  37
  38   private
  39
  40   def parse_and_transform
  41     return if @source_html.blank?
  42
  43     parsed_html = Nokogiri::HTML.fragment(@source_html)
  44
  45     parsed_html.traverse do |node|
  46       next unless TARGET_ELEMENTS.include?(node.name)
  47
  48       anchor = node['id'] || node.text.parameterize
  49       @slugs[anchor] += 1
  50       anchor = "#{anchor}-#{@slugs[anchor]}" if @slugs[anchor] > 1
  51
  52       node['id'] = anchor
  53
  54       next unless LISTED_ELEMENTS.include?(node.name)
  55
  56       depth          = node.name[1..-1]
  57       latest_section = @headers.last
  58
  59       if latest_section.nil? || latest_section.depth >= depth
  60         @headers << Section.new(depth, node.text, anchor)
  61       else
  62         latest_section << Section.new(depth, node.text, anchor)
  63       end
  64     end
  65
  66     @target_html = parsed_html.to_s
  67     @processed   = true
  68   end
  69 end