Refactor TableOfContentsFilter's nested table of contents

Most of the logic is now self-contained within the `HeaderNode` class.

1 files changed, 64 insertions, 42 deletions

diff --git a/lib/banzai/filter/table_of_contents_filter.rb b/lib/banzai/filter/table_of_contents_filter.rb
index 8cb860c5a92..eda94bea5e0 100644
--- a/lib/banzai/filter/table_of_contents_filter.rb
+++ b/lib/banzai/filter/table_of_contents_filter.rb
@@ -15,7 +15,6 @@ module Banzai
     #          `li` child elements.
     class TableOfContentsFilter < HTML::Pipeline::Filter
       PUNCTUATION_REGEXP = /[^\p{Word}\- ]/u
-      HeaderNode = Struct.new(:level, :href, :text, :children, :parent)
 
       def call
         return doc if context[:no_header_anchors]
@@ -23,56 +22,34 @@ module Banzai
         result[:toc] = ""
 
         headers = Hash.new(0)
-
-        # root node of header-tree
-        header_root = HeaderNode.new(0, nil, nil, [], nil)
-        current_header = header_root
+        header_root = current_header = HeaderNode.new
 
         doc.css('h1, h2, h3, h4, h5, h6').each do |node|
-          text = node.text
-
-          id = text.downcase
-          id.gsub!(PUNCTUATION_REGEXP, '') # remove punctuation
-          id.tr!(' ', '-') # replace spaces with dash
-          id.squeeze!('-') # replace multiple dashes with one
-
-          uniq = (headers[id] > 0) ? "-#{headers[id]}" : ''
-          headers[id] += 1
-
           if header_content = node.children.first
-            # namespace detection will be automatically handled via javascript (see issue #22781)
-            namespace = "user-content-"
+            id = node
+              .text
+              .downcase
+              .gsub(PUNCTUATION_REGEXP, '') # remove punctuation
+              .tr(' ', '-') # replace spaces with dash
+              .squeeze('-') # replace multiple dashes with one
+
+            uniq = headers[id] > 0 ? "-#{headers[id]}" : ''
+            headers[id] += 1
             href = "#{id}#{uniq}"
 
-            level = node.name[1].to_i # get this header level
-            if level == current_header.level
-              # same as previous
-              parent = current_header.parent
-            elsif level > current_header.level
-              # larger (weaker) than previous
-              parent = current_header
-            else
-              # smaller (stronger) than previous
-              # search parent
-              parent = current_header
-              parent = parent.parent while parent.level >= level
-            end
-
-            # create header-node and push as child
-            header_node = HeaderNode.new(level, href, text, [], parent)
-            parent.children.push(header_node)
-            current_header = header_node
-
-            header_content.add_previous_sibling(anchor_tag("#{namespace}#{href}", href))
+            current_header = HeaderNode.new(node: node, href: href, previous_header: current_header)
+
+            header_content.add_previous_sibling(anchor_tag(href))
           end
         end
 
-        # extract header-tree
         if header_root.children.length > 0
-          result[:toc] = %Q{<ul class="section-nav">\n}
+          result[:toc] = %q{<ul class="section-nav">}
+
           header_root.children.each do |child|
             push_toc(child)
           end
+
           result[:toc] << '</ul>'
         end
 
@@ -81,20 +58,65 @@ module Banzai
 
       private
 
-      def anchor_tag(id, href)
-        %Q{<a id="#{id}" class="anchor" href="##{href}" aria-hidden="true"></a>}
+      def anchor_tag(href)
+        %Q{<a id="user-content-#{href}" class="anchor" href="##{href}" aria-hidden="true"></a>}
       end
 
       def push_toc(header_node)
         result[:toc] << %Q{<li><a href="##{header_node.href}">#{header_node.text}</a>}
+
         if header_node.children.length > 0
           result[:toc] << '<ul>'
+
           header_node.children.each do |child|
             push_toc(child)
           end
+
           result[:toc] << '</ul>'
         end
-        result[:toc] << '</li>\n'
+
+        result[:toc] << '</li>'
+      end
+
+      class HeaderNode
+        attr_reader :node, :href, :parent, :children
+
+        def initialize(node: nil, href: nil, previous_header: nil)
+          @node = node
+          @href = href
+          @children = []
+
+          find_parent(previous_header)
+        end
+
+        def level
+          return 0 unless node
+
+          @level ||= node.name[1].to_i
+        end
+
+        def text
+          return '' unless node
+
+          @text ||= node.text
+        end
+
+        private
+
+        def find_parent(previous_header)
+          return unless previous_header
+
+          if level == previous_header.level
+            @parent = previous_header.parent
+          elsif level > previous_header.level
+            @parent = previous_header
+          else
+            @parent = previous_header
+            @parent = @parent.parent while @parent.level >= level
+          end
+
+          @parent.children.push(self)
+        end
       end
     end
   end