From 34ab6dfa051c29d27353a9f555e713f36c7954a4 Mon Sep 17 00:00:00 2001
From: Brett Walker <bwalker@gitlab.com>
Date: Fri, 4 Jan 2019 20:54:02 -0600
Subject: Properly process footnotes in markdown

All the ids and classes were stripped.  Add them back in
and make ids unique
---
 lib/banzai/filter/footnote_filter.rb     | 55 ++++++++++++++++++++++++++++++++
 lib/banzai/filter/sanitization_filter.rb | 27 ++++++++++++++--
 lib/banzai/pipeline/gfm_pipeline.rb      |  1 +
 3 files changed, 81 insertions(+), 2 deletions(-)
 create mode 100644 lib/banzai/filter/footnote_filter.rb

(limited to 'lib/banzai')

diff --git a/lib/banzai/filter/footnote_filter.rb b/lib/banzai/filter/footnote_filter.rb
new file mode 100644
index 00000000000..a7120fbd46e
--- /dev/null
+++ b/lib/banzai/filter/footnote_filter.rb
@@ -0,0 +1,55 @@
+# frozen_string_literal: true
+
+module Banzai
+  module Filter
+    # HTML Filter for footnotes
+    #
+    # Footnotes are supported in CommonMark.  However we were stripping
+    # the ids during sanitization.  Those are now allowed.
+    #
+    # Footnotes are numbered the same - the first one has `id=fn1`, the
+    # second is `id=fn2`, etc.  In order to allow footnotes when rendering
+    # multiple markdown blocks on a page, we need to make each footnote
+    # reference unique.
+    #
+    # This filter adds a random number to each footnote (the same number
+    # can be used for a single render). So you get `id=fn1-4335` and `id=fn2-4335`.
+    #
+    class FootnoteFilter < HTML::Pipeline::Filter
+      INTEGER_PATTERN = /\A\d+\Z/.freeze
+
+      def call
+        return doc unless first_footnote = doc.at_css('ol > li[id=fn1]')
+
+        # Sanitization stripped off the section wrapper - add it back in
+        first_footnote.parent.wrap('<section class="footnotes">')
+
+        doc.css('sup > a[id]').each do |link_node|
+          ref_num       = link_node[:id].delete_prefix('fnref')
+          footnote_node = doc.at_css("li[id=fn#{ref_num}]")
+          backref_node  = doc.at_css("li[id=fn#{ref_num}] a[href=\"#fnref#{ref_num}\"]")
+
+          if ref_num =~ INTEGER_PATTERN && footnote_node && backref_node
+            rand_ref_num        = "#{ref_num}-#{random_number}"
+            link_node[:href]    = "#fn#{rand_ref_num}"
+            link_node[:id]      = "fnref#{rand_ref_num}"
+            footnote_node[:id]  = "fn#{rand_ref_num}"
+            backref_node[:href] = "#fnref#{rand_ref_num}"
+
+            # Sanitization stripped off class - add it back in
+            link_node.parent.append_class('footnote-ref')
+            backref_node.append_class('footnote-backref')
+          end
+        end
+
+        doc
+      end
+
+      private
+
+      def random_number
+        @random_number ||= rand(10000)
+      end
+    end
+  end
+end
diff --git a/lib/banzai/filter/sanitization_filter.rb b/lib/banzai/filter/sanitization_filter.rb
index 8ba09290e6d..d05518edcea 100644
--- a/lib/banzai/filter/sanitization_filter.rb
+++ b/lib/banzai/filter/sanitization_filter.rb
@@ -8,8 +8,10 @@ module Banzai
     class SanitizationFilter < HTML::Pipeline::SanitizationFilter
       include Gitlab::Utils::StrongMemoize
 
-      UNSAFE_PROTOCOLS = %w(data javascript vbscript).freeze
-      TABLE_ALIGNMENT_PATTERN = /text-align: (?<alignment>center|left|right)/
+      UNSAFE_PROTOCOLS                = %w(data javascript vbscript).freeze
+      TABLE_ALIGNMENT_PATTERN         = /text-align: (?<alignment>center|left|right)/.freeze
+      FOOTNOTE_LINK_REFERENCE_PATTERN = /\Afnref\d\z/.freeze
+      FOOTNOTE_LI_REFERENCE_PATTERN   = /\Afn\d\z/.freeze
 
       def whitelist
         strong_memoize(:whitelist) do
@@ -57,6 +59,13 @@ module Banzai
         # Remove any `style` properties not required for table alignment
         whitelist[:transformers].push(self.class.remove_unsafe_table_style)
 
+        # Allow `id` in a and li elements for footnotes
+        whitelist[:attributes]['a'].push('id')
+        whitelist[:attributes]['li'] = %w(id)
+
+        # ...but remove any `id` properties not matching for footnotes
+        whitelist[:transformers].push(self.class.remove_non_footnote_ids)
+
         whitelist
       end
 
@@ -112,6 +121,20 @@ module Banzai
             end
           end
         end
+
+        def remove_non_footnote_ids
+          lambda do |env|
+            node = env[:node]
+
+            return unless node.name == 'a' || node.name == 'li'
+            return unless node.has_attribute?('id')
+
+            return if node.name == 'a' && node['id'] =~ FOOTNOTE_LINK_REFERENCE_PATTERN
+            return if node.name == 'li' && node['id'] =~ FOOTNOTE_LI_REFERENCE_PATTERN
+
+            node.remove_attribute('id')
+          end
+        end
       end
     end
   end
diff --git a/lib/banzai/pipeline/gfm_pipeline.rb b/lib/banzai/pipeline/gfm_pipeline.rb
index 5f13a6d6cde..d860dad0b6c 100644
--- a/lib/banzai/pipeline/gfm_pipeline.rb
+++ b/lib/banzai/pipeline/gfm_pipeline.rb
@@ -30,6 +30,7 @@ module Banzai
           Filter::AutolinkFilter,
           Filter::ExternalLinkFilter,
           Filter::SuggestionFilter,
+          Filter::FootnoteFilter,
 
           *reference_filters,
 
-- 
cgit v1.2.3