X-Git-Url: https://git.openstreetmap.org./rails.git/blobdiff_plain/9a9b045372a6f48420a9a6dacfde52c34ab7abce..70cb93a101638596ecae5782575ada2d4d08c8dd:/lib/rich_text.rb diff --git a/lib/rich_text.rb b/lib/rich_text.rb index 7325a2a28..a439342f7 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 +29,32 @@ 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 + + spammy_phrases = SPAMMY_PHRASES.count do |phrase| + doc.content.include?(phrase) end - [link_proportion - 0.2, 0.0].max * 200 + link_count * 40 + ([link_proportion - 0.2, 0.0].max * 200) + + (link_count * 40) + + (spammy_phrases * 40) + end + + def image + nil + end + + def image_alt + nil end protected @@ -45,11 +63,15 @@ 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 @@ -62,44 +84,45 @@ 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 - Markdown.html_parser.render(self).html_safe + linkify(sanitize(document.to_html), :all) end def to_text to_s end - def self.html_renderer - @html_renderer ||= Renderer.new(:filter_html => true, :safe_links_only => true) + def image + @image_element = first_image_element(document.root) unless defined? @image_element + @image_element.attr["src"] if @image_element end - def self.html_parser - @html_parser ||= Redcarpet::Markdown.new(html_renderer, :no_intra_emphasis => true, :autolink => true, :space_after_headers => true) + def image_alt + @image_element = first_image_element(document.root) unless defined? @image_element + @image_element.attr["alt"] if @image_element end - class Renderer < Redcarpet::Render::XHTML - def link(link, _title, alt_text) - "#{alt_text}" - end + private - def autolink(link, link_type) - if link_type == :email - "#{link}" - else - "#{link}" - end + def document + @document ||= Kramdown::Document.new(self) + end + + def first_image_element(element) + return element if image?(element) && element.attr["src"].present? + + element.children.find do |child| + nested_image = first_image_element(child) + break nested_image if nested_image end end + + def image?(element) + element.type == :img || (element.type == :html_element && element.value == "img") + end end class Text < Base