Add Gitlab::Markdown::TableOfContentsFilter

Removes header and table of contents processing from Redcarpet renderer.

3 files changed, 67 insertions, 10 deletions

diff --git a/lib/gitlab/markdown.rb b/lib/gitlab/markdown.rb
index 44779d7fdd8..ffb7e16aed2 100644
--- a/lib/gitlab/markdown.rb
+++ b/lib/gitlab/markdown.rb
@@ -38,6 +38,7 @@ module Gitlab
     autoload :LabelReferenceFilter,         'gitlab/markdown/label_reference_filter'
     autoload :MergeRequestReferenceFilter,  'gitlab/markdown/merge_request_reference_filter'
     autoload :SnippetReferenceFilter,       'gitlab/markdown/snippet_reference_filter'
+    autoload :TableOfContentsFilter,        'gitlab/markdown/table_of_contents_filter'
     autoload :UserReferenceFilter,          'gitlab/markdown/user_reference_filter'
 
     # Public: Parse the provided text with GitLab-Flavored Markdown
@@ -81,6 +82,9 @@ module Gitlab
         asset_root: Gitlab.config.gitlab.url,
         asset_host: Gitlab::Application.config.asset_host,
 
+        # TableOfContentsFilter
+        no_header_anchors: options[:no_header_anchors],
+
         # ReferenceFilter
         current_user:    current_user,
         only_path:       options[:reference_only_path],
@@ -117,6 +121,7 @@ module Gitlab
         HTML::Pipeline::SanitizationFilter,
 
         Gitlab::Markdown::EmojiFilter,
+        Gitlab::Markdown::TableOfContentsFilter,
 
         Gitlab::Markdown::UserReferenceFilter,
         Gitlab::Markdown::IssueReferenceFilter,
diff --git a/lib/gitlab/markdown/table_of_contents_filter.rb b/lib/gitlab/markdown/table_of_contents_filter.rb
new file mode 100644
index 00000000000..c97612dafb8
--- /dev/null
+++ b/lib/gitlab/markdown/table_of_contents_filter.rb
@@ -0,0 +1,62 @@
+require 'html/pipeline/filter'
+
+module Gitlab
+  module Markdown
+    # HTML filter that adds an anchor child element to all Headers in a
+    # document, so that they can be linked to.
+    #
+    # Generates the Table of Contents with links to each header. See Results.
+    #
+    # Based on HTML::Pipeline::TableOfContentsFilter.
+    #
+    # Context options:
+    #   :no_header_anchors - Skips all processing done by this filter.
+    #
+    # Results:
+    #   :toc - String containing Table of Contents data as a `ul` element with
+    #          `li` child elements.
+    class TableOfContentsFilter < HTML::Pipeline::Filter
+      PUNCTUATION_REGEXP = /[^\p{Word}\- ]/u
+
+      def call
+        return doc if context[:no_header_anchors]
+
+        result[:toc] = ""
+
+        headers = Hash.new(0)
+
+        doc.css('h1, h2, h3, h4, h5, h6').each do |node|
+          text = node.text
+
+          id = text.downcase
+          id.gsub!(PUNCTUATION_REGEXP, '') # remove punctuation
+          id.gsub!(' ', '-') # replace spaces with dash
+          id.squeeze!(' -') # replace multiple spaces or dashes with one
+
+          uniq = (headers[id] > 0) ? "-#{headers[id]}" : ''
+          headers[id] += 1
+
+          if header_content = node.children.first
+            href = "#{id}#{uniq}"
+            push_toc(href, text)
+            header_content.add_previous_sibling(anchor_tag(href))
+          end
+        end
+
+        result[:toc] = %Q{<ul class="section-nav">\n#{result[:toc]}</ul>} unless result[:toc].empty?
+
+        doc
+      end
+
+      private
+
+      def anchor_tag(href)
+        %Q{<a id="#{href}" class="anchor" href="##{href}" aria-hidden="true"></a>}
+      end
+
+      def push_toc(href, text)
+        result[:toc] << %Q{<li><a href="##{href}">#{text}</a></li>\n}
+      end
+    end
+  end
+end
diff --git a/lib/redcarpet/render/gitlab_html.rb b/lib/redcarpet/render/gitlab_html.rb
index 10efff2ae9f..dc5fbe3c8e1 100644
--- a/lib/redcarpet/render/gitlab_html.rb
+++ b/lib/redcarpet/render/gitlab_html.rb
@@ -50,16 +50,6 @@ class Redcarpet::Render::GitlabHTML < Redcarpet::Render::HTML
     h.link_to_gfm(content, link, title: title)
   end
 
-  def header(text, level)
-    if @options[:no_header_anchors]
-      "<h#{level}>#{text}</h#{level}>"
-    else
-      id = ActionController::Base.helpers.strip_tags(h.gfm(text)).downcase() \
-          .gsub(/[^a-z0-9_-]/, '-').gsub(/-+/, '-').gsub(/^-/, '').gsub(/-$/, '')
-      "<h#{level} id=\"#{id}\">#{text}<a href=\"\##{id}\"></a></h#{level}>"
-    end
-  end
-
   def postprocess(full_document)
     full_document.gsub!("ftp://smb:", "smb://")