X-Git-Url: https://git.openstreetmap.org./rails.git/blobdiff_plain/5cbd4038edb32b0304bd766e70fc680ea447b52b..24c138ac8d5952117611146cec78379eb12e315f:/lib/rich_text.rb diff --git a/lib/rich_text.rb b/lib/rich_text.rb index bb2baddc8..f19d3d3a9 100644 --- a/lib/rich_text.rb +++ b/lib/rich_text.rb @@ -1,4 +1,8 @@ module RichText + SPAMMY_PHRASES = [ + "Business Description:", "Additional Keywords:" + ].freeze + def self.new(format, text) case format when "html" then HTML.new(text || "") @@ -11,7 +15,7 @@ module RichText include ActionView::Helpers::TextHelper include ActionView::Helpers::OutputSafetyHelper - def sanitize(text) + def sanitize(text, _options = {}) Sanitize.clean(text, Sanitize::Config::OSM).html_safe end end @@ -25,18 +29,28 @@ module RichText doc = Nokogiri::HTML(to_html) - if doc.content.length > 0 + if doc.content.empty? + link_proportion = 0 + else doc.xpath("//a").each do |link| link_count += 1 link_size += link.content.length end - link_proportion = link_size.to_f / doc.content.length.to_f - else - link_proportion = 0 + link_proportion = link_size.to_f / doc.content.length end - [link_proportion - 0.2, 0.0].max * 200 + link_count * 40 + spammy_phrases = SPAMMY_PHRASES.count do |phrase| + doc.content.include?(phrase) + end + + ([link_proportion - 0.2, 0.0].max * 200) + + (link_count * 40) + + (spammy_phrases * 40) + end + + def image + nil end protected @@ -45,11 +59,15 @@ module RichText SimpleFormat.new.simple_format(text) end - def linkify(text) + def sanitize(text) + Sanitize.clean(text, Sanitize::Config::OSM).html_safe + end + + def linkify(text, mode = :urls) if text.html_safe? - Rinku.auto_link(text, :urls, tag_options(:rel => "nofollow")).html_safe + Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")).html_safe else - Rinku.auto_link(text, :urls, tag_options(:rel => "nofollow")) + Rinku.auto_link(text, mode, tag_builder.tag_options(:rel => "nofollow noopener noreferrer")) end end end @@ -62,41 +80,35 @@ module RichText def to_text to_s end - - private - - def sanitize(text) - Sanitize.clean(text, Sanitize::Config::OSM).html_safe - end end class Markdown < Base def to_html - html_parser.render(self).html_safe + linkify(sanitize(document.to_html), :all) end def to_text to_s end + def image + return @image if defined? @image + + @image = first_image_element(document.root)&.attr&.[]("src") + end + private - def html_parser - @@html_renderer ||= Renderer.new(:filter_html => true, :safe_links_only => true) - @@html_parser ||= Redcarpet::Markdown.new(@@html_renderer, :no_intra_emphasis => true, :autolink => true, :space_after_headers => true) + def document + @document ||= Kramdown::Document.new(self) end - class Renderer < Redcarpet::Render::XHTML - def link(link, _title, alt_text) - "#{alt_text}" - end + def first_image_element(element) + return element if element.type == :img - def autolink(link, link_type) - if link_type == :email - "#{link}" - else - "#{link}" - end + element.children.find do |child| + nested_image = first_image_element(child) + break nested_image if nested_image end end end