Preserve footnote link ids

3 files changed, 83 insertions, 5 deletions

diff --git a/changelogs/unreleased/64645-asciidoctor-preserve-footnote-link-ids.yml b/changelogs/unreleased/64645-asciidoctor-preserve-footnote-link-ids.yml
new file mode 100644
index 00000000000..5427a035478
--- /dev/null
+++ b/changelogs/unreleased/64645-asciidoctor-preserve-footnote-link-ids.yml
@@ -0,0 +1,5 @@
+---
+title: "Preserve footnote link ids in Asciidoctor"
+merge_request: 30790
+author: Guillaume Grossetie
+type: added
\ No newline at end of file
diff --git a/lib/banzai/filter/ascii_doc_sanitization_filter.rb b/lib/banzai/filter/ascii_doc_sanitization_filter.rb
index d8d63075752..9105e86ad04 100644
--- a/lib/banzai/filter/ascii_doc_sanitization_filter.rb
+++ b/lib/banzai/filter/ascii_doc_sanitization_filter.rb
@@ -8,12 +8,18 @@ module Banzai
     class AsciiDocSanitizationFilter < Banzai::Filter::BaseSanitizationFilter
       # Section anchor link pattern
       SECTION_LINK_REF_PATTERN = /\A#{Gitlab::Asciidoc::DEFAULT_ADOC_ATTRS['idprefix']}(:?[[:alnum:]]|-|_)+\z/.freeze
+      SECTION_HEADINGS = %w(h2 h3 h4 h5 h6).freeze
+
+      # Footnote link patterns
+      FOOTNOTE_LINK_ID_PATTERNS = {
+        a: /\A_footnoteref_\d+\z/,
+        div: /\A_footnotedef_\d+\z/
+      }.freeze
 
       # Classes used by Asciidoctor to style components
       ADMONITION_CLASSES = %w(fa icon-note icon-tip icon-warning icon-caution icon-important).freeze
       CALLOUT_CLASSES = ['conum'].freeze
       CHECKLIST_CLASSES = %w(fa fa-check-square-o fa-square-o).freeze
-
       LIST_CLASSES = %w(checklist none no-bullet unnumbered unstyled).freeze
 
       ELEMENT_CLASSES_WHITELIST = {
@@ -26,8 +32,6 @@ module Banzai
         a: ['anchor'].freeze
       }.freeze
 
-      ALLOWED_HEADERS = %w(h2 h3 h4 h5 h6).freeze
-
       def customize_whitelist(whitelist)
         # Allow marks
         whitelist[:elements].push('mark')
@@ -44,20 +48,39 @@ module Banzai
         whitelist[:transformers].push(self.class.remove_element_classes)
 
         # Allow `id` in heading elements for section anchors
-        ALLOWED_HEADERS.each do |header|
+        SECTION_HEADINGS.each do |header|
           whitelist[:attributes][header] = %w(id)
         end
         whitelist[:transformers].push(self.class.remove_non_heading_ids)
 
+        # Allow `id` in footnote elements
+        FOOTNOTE_LINK_ID_PATTERNS.keys.each do |element|
+          whitelist[:attributes][element.to_s].push('id')
+        end
+        whitelist[:transformers].push(self.class.remove_non_footnote_ids)
+
         whitelist
       end
 
       class << self
+        def remove_non_footnote_ids
+          lambda do |env|
+            node = env[:node]
+
+            return unless (pattern = FOOTNOTE_LINK_ID_PATTERNS[node.name.to_sym])
+            return unless node.has_attribute?('id')
+
+            return if node['id'] =~ pattern
+
+            node.remove_attribute('id')
+          end
+        end
+
         def remove_non_heading_ids
           lambda do |env|
             node = env[:node]
 
-            return unless ALLOWED_HEADERS.any?(node.name)
+            return unless SECTION_HEADINGS.any?(node.name)
             return unless node.has_attribute?('id')
 
             return if node['id'] =~ SECTION_LINK_REF_PATTERN
diff --git a/spec/lib/gitlab/asciidoc_spec.rb b/spec/lib/gitlab/asciidoc_spec.rb
index 5293732ead1..cbd4a509a55 100644
--- a/spec/lib/gitlab/asciidoc_spec.rb
+++ b/spec/lib/gitlab/asciidoc_spec.rb
@@ -110,6 +110,56 @@ module Gitlab
 
           expect(render(input, context)).to include(output.strip)
         end
+
+        it 'removes non footnote def ids' do
+          input = <<~ADOC
+            ++++
+            <div id="def">Footnote definition</div>
+            ++++
+          ADOC
+
+          output = <<~HTML
+            <div>Footnote definition</div>
+          HTML
+
+          expect(render(input, context)).to include(output.strip)
+        end
+
+        it 'removes non footnote ref ids' do
+          input = <<~ADOC
+            ++++
+            <a id="ref">Footnote reference</a>
+            ++++
+          ADOC
+
+          output = <<~HTML
+            <a>Footnote reference</a>
+          HTML
+
+          expect(render(input, context)).to include(output.strip)
+        end
+      end
+
+      context 'with footnotes' do
+        it 'preserves ids and links' do
+          input = <<~ADOC
+            This paragraph has a footnote.footnote:[This is the text of the footnote.]
+          ADOC
+
+          output = <<~HTML
+           <div>
+           <p>This paragraph has a footnote.<sup>[<a id="_footnoteref_1" href="#_footnotedef_1" title="View footnote.">1</a>]</sup></p>
+           </div>
+           <div>
+           <hr>
+           <div id="_footnotedef_1">
+           <a href="#_footnoteref_1">1</a>. This is the text of the footnote.
+           </div>
+           </div>
+          HTML
+
+          expect(render(input, context)).to include(output.strip)
+        end
       end
 
       context 'with section anchors' do