4 files changed, 33 insertions, 12 deletions

diff --git a/lib/gitlab/docs/element.rb b/lib/gitlab/docs/element.rb
index a78b9a01..4d58edee 100644
--- a/lib/gitlab/docs/element.rb
+++ b/lib/gitlab/docs/element.rb
@@ -1,3 +1,5 @@
+require 'cgi'
+
 module Gitlab
   module Docs
     class Element
@@ -15,17 +17,25 @@ module Gitlab
       end
 
       def href
-        @href ||= attribute('href')
+        @href ||= self.class.decode(attribute('href'))
       end
 
       def id
-        @id ||= attribute('id')
+        @id ||= attribute('id')&.downcase
+      end
+
+      def self.decode(name)
+        return if name.to_s.empty?
+
+        CGI.unescape(name)
       end
 
       private
 
       def attribute(name)
-        @attributes.find { |attr| attr.first == name }&.last
+        @attributes.to_a
+          .find { |attr| attr.first == name }
+          .to_a.last
       end
     end
   end
diff --git a/lib/gitlab/docs/link.rb b/lib/gitlab/docs/link.rb
index 059ebaf1..23460f88 100644
--- a/lib/gitlab/docs/link.rb
+++ b/lib/gitlab/docs/link.rb
@@ -15,7 +15,7 @@ module Gitlab
       def anchor_name
         return unless @href.include?('#')
 
-        @href.partition('#').last.downcase
+        @href.partition('#').last
       end
 
       def internal_anchor?
diff --git a/lib/gitlab/docs/page.rb b/lib/gitlab/docs/page.rb
index 8bd5e9b9..463e3f36 100644
--- a/lib/gitlab/docs/page.rb
+++ b/lib/gitlab/docs/page.rb
@@ -6,8 +6,8 @@ module Gitlab
 
       def initialize(file)
         @file = file
-        @hrefs = []
-        @ids = []
+        @hrefs = Set.new
+        @ids = Set.new
 
         return unless exists?
 
@@ -31,7 +31,7 @@ module Gitlab
       end
 
       def has_anchor?(name)
-        @ids.include?(name)
+        @ids.include?(Docs::Element.decode(name.downcase))
       end
 
       def self.build(path)
diff --git a/spec/gitlab/docs/page_spec.rb b/spec/gitlab/docs/page_spec.rb
index 64cc98da..a9ee3264 100644
--- a/spec/gitlab/docs/page_spec.rb
+++ b/spec/gitlab/docs/page_spec.rb
@@ -25,8 +25,11 @@ describe Gitlab::Docs::Page do
               <a href="../link.html#my-anchor">See external file</a>
               <a href="#internal-anchor">See internal section</a>
               <a href="#internal-anchor-2">See internal section broken</a>
+              <a href="#utf-8-id-✔">See utf-8 anchor</a>
+              <a href="#utf-8-id-%E2%9C%94">See utf-8 anchor 2</a>
 
               <h1 id="internal-anchor">Some section</h1>
+              <h1 id="utf-8-id-✔">Some section 2</h1>
             </body
           </html>
         HTML
@@ -34,7 +37,7 @@ describe Gitlab::Docs::Page do
 
     describe '#links' do
       it 'collects links on a page' do
-        expect(subject.links.count).to eq 3
+        expect(subject.links.count).to eq 4
       end
     end
 
@@ -42,23 +45,31 @@ describe Gitlab::Docs::Page do
       it 'collects all hrefs' do
         expect(subject.hrefs).to match_array %w[../link.html#my-anchor
                                                 #internal-anchor
-                                                #internal-anchor-2]
+                                                #internal-anchor-2
+                                                #utf-8-id-✔]
       end
     end
 
     describe '#ids' do
       it 'collects all ids' do
-        expect(subject.ids).to match_array %w[internal-anchor]
+        expect(subject.ids).to match_array %w[internal-anchor
+                                              utf-8-id-✔]
       end
     end
 
     describe '#has_anchor?' do
       it 'returns true when anchor exists on a page' do
-        expect(subject.has_anchor?('internal-anchor')).to be true
+        expect(subject).to have_anchor('internal-anchor')
       end
 
       it 'returns false when anchors does not exist' do
-        expect(subject.has_anchor?('internal-anchor-2')).to be false
+        expect(subject).not_to have_anchor('internal-anchor-2')
+      end
+
+      it 'returns true when UTF-8 encoded anchors are compared' do
+        expect(subject).to have_anchor('utf-8-id-✔')
+        expect(subject).to have_anchor('utf-8-id-%E2%9C%94')
+        expect(subject).to have_anchor('utf-8-id-%e2%9c%94')
       end
     end
   end