lib/spam_scorer/rich_text.rb

   1 # frozen_string_literal: true
   2
   3 module SpamScorer
   4   class RichText
   5     def initialize(text)
   6       @text = text
   7     end
   8
   9     def score
  10       link_count = 0
  11       link_size = 0
  12
  13       doc = Nokogiri::HTML(text.to_html)
  14
  15       if doc.content.empty?
  16         link_proportion = 0
  17       else
  18         doc.xpath("//a").each do |link|
  19           link_count += 1
  20           link_size += link.content.length
  21         end
  22
  23         link_proportion = link_size.to_f / doc.content.length
  24       end
  25
  26       comparable_content = to_comparable_form(doc.content)
  27       spammy_phrases = SpammyPhrase.pluck(:phrase).count do |phrase|
  28         comparable_content.include?(to_comparable_form(phrase))
  29       end
  30
  31       ([link_proportion - 0.2, 0.0].max * 200) +
  32         (link_count * 40) +
  33         (spammy_phrases * 40)
  34     end
  35
  36     private
  37
  38     attr_reader :text
  39
  40     def to_comparable_form(str)
  41       str.downcase(:fold).unicode_normalize(:nfkc)
  42     end
  43   end
  44 end