X-Git-Url: https://git.openstreetmap.org./rails.git/blobdiff_plain/aa4205482a4af317ec26fc686793b57f02a6864a..70cb93a101638596ecae5782575ada2d4d08c8dd:/lib/rich_text.rb diff --git a/lib/rich_text.rb b/lib/rich_text.rb index ec5e9e473..a439342f7 100644 --- a/lib/rich_text.rb +++ b/lib/rich_text.rb @@ -1,57 +1,137 @@ module RichText + SPAMMY_PHRASES = [ + "Business Description:", "Additional Keywords:" + ].freeze + def self.new(format, text) case format - when "html"; HTML.new(text || "") - when "markdown"; Markdown.new(text || "") - else; nil + when "html" then HTML.new(text || "") + when "markdown" then Markdown.new(text || "") + when "text" then Text.new(text || "") end end - class HTML < String + class SimpleFormat include ActionView::Helpers::TextHelper + include ActionView::Helpers::OutputSafetyHelper + + def sanitize(text, _options = {}) + Sanitize.clean(text, Sanitize::Config::OSM).html_safe + end + end + + class Base < String include ActionView::Helpers::TagHelper - def to_html - linkify(sanitize(simple_format(self))) + def spam_score + link_count = 0 + link_size = 0 + + doc = Nokogiri::HTML(to_html) + + 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 + end + + 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 to_text - self + def image + nil end - private + def image_alt + nil + end + + protected + + def simple_format(text) + SimpleFormat.new.simple_format(text) + end def sanitize(text) Sanitize.clean(text, Sanitize::Config::OSM).html_safe end - def linkify(text) + 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 - class Markdown < String + class HTML < Base + def to_html + linkify(sanitize(simple_format(self))) + end + + def to_text + to_s + end + end + + class Markdown < Base def to_html - html_parser.render(self).html_safe + linkify(sanitize(document.to_html), :all) end def to_text - self + to_s end - private + def image + @image_element = first_image_element(document.root) unless defined? @image_element + @image_element.attr["src"] if @image_element + end + + def image_alt + @image_element = first_image_element(document.root) unless defined? @image_element + @image_element.attr["alt"] if @image_element + end + + private + + def document + @document ||= Kramdown::Document.new(self) + end - def html_parser - @@html_renderer ||= Redcarpet::Render::XHTML.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 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 + def to_html + linkify(simple_format(ERB::Util.html_escape(self))) + end + + def to_text + to_s end end end