From 3ea5b948a4cee9ea5a1e229f567974c323947ef5 Mon Sep 17 00:00:00 2001
From: Yamagishi Kazutoshi <ykzts@desire.sh>
Date: Tue, 25 Apr 2017 22:03:51 +0900
Subject: [PATCH] Decodes URL containing IDN (#2436)

---
 app/lib/formatter.rb       | 10 +++-------
 spec/lib/formatter_spec.rb | 21 ++++++++++++++++-----
 2 files changed, 19 insertions(+), 12 deletions(-)

diff --git a/app/lib/formatter.rb b/app/lib/formatter.rb
index 12e79c9b1..1d8e90d1f 100644
--- a/app/lib/formatter.rb
+++ b/app/lib/formatter.rb
@@ -61,20 +61,15 @@ class Formatter
     result = ''
 
     last_index = entities.reduce(0) do |index, entity|
+      normalized_url = Addressable::URI.parse(entity[:url]).normalize
       indices = entity[:indices]
       result += encode(chars[index...indices.first].join)
-      result += Twitter::Autolink.send(:link_to_text, entity, link_html(entity[:url]), entity[:url], html_attrs)
+      result += Twitter::Autolink.send(:link_to_text, entity, link_html(entity[:url]), normalized_url, html_attrs)
       indices.last
     end
     result += encode(chars[last_index..-1].join)
   end
 
-  def link_urls(html)
-    Twitter::Autolink.auto_link_urls(html, url_target: '_blank',
-                                           link_attribute_block: lambda { |_, a| a[:rel] << ' noopener' },
-                                           link_text_block: lambda { |_, text| link_html(text) })
-  end
-
   def link_mentions(html, mentions)
     html.gsub(Account::MENTION_RE) do |match|
       acct    = Account::MENTION_RE.match(match)[1]
@@ -102,6 +97,7 @@ class Formatter
   end
 
   def link_html(url)
+    url = Addressable::URI.parse(url).display_uri.to_s
     prefix = url.match(/\Ahttps?:\/\/(www\.)?/).to_s
     text   = url[prefix.length, 30]
     suffix = url[prefix.length + 30..-1]
diff --git a/spec/lib/formatter_spec.rb b/spec/lib/formatter_spec.rb
index b70231d26..b762907b2 100644
--- a/spec/lib/formatter_spec.rb
+++ b/spec/lib/formatter_spec.rb
@@ -18,7 +18,7 @@ RSpec.describe Formatter do
     end
 
     it 'contains a link' do
-      expect(subject).to match('<a href="http://google.com" rel="nofollow noopener" target="_blank"><span class="invisible">http://</span><span class="">google.com</span><span class="invisible"></span></a>')
+      expect(subject).to match('<a href="http://google.com/" rel="nofollow noopener" target="_blank"><span class="invisible">http://</span><span class="">google.com/</span><span class="invisible"></span></a>')
     end
 
     context 'matches a stand-alone medium URL' do
@@ -31,7 +31,18 @@ RSpec.describe Formatter do
     context 'matches a stand-alone google URL' do
       let(:local_text) { 'http://google.com' }
       it 'has valid url' do
-        expect(subject).to include('href="http://google.com"')
+        expect(subject).to include('href="http://google.com/"')
+      end
+    end
+
+    context 'matches a stand-alone IDN URL' do
+      let(:local_text) { 'https://nic.みんな/' }
+      it 'has valid url' do
+        expect(subject).to include('href="https://nic.xn--q9jyb4c/"')
+      end
+
+      it 'has display url' do
+        expect(subject).to include('<span class="">nic.みんな/</span>')
       end
     end
 
@@ -51,21 +62,21 @@ RSpec.describe Formatter do
     context 'matches a URL without exclamation point' do
       let(:local_text) { 'http://www.google.com!' }
       it 'has valid url' do
-        expect(subject).to include('href="http://www.google.com"')
+        expect(subject).to include('href="http://www.google.com/"')
       end
     end
 
     context 'matches a URL without single quote' do
       let(:local_text) { "http://www.google.com'" }
       it 'has valid url' do
-        expect(subject).to include('href="http://www.google.com"')
+        expect(subject).to include('href="http://www.google.com/"')
       end
     end
 
     context 'matches a URL without angle brackets' do
       let(:local_text) { 'http://www.google.com>' }
       it 'has valid url' do
-        expect(subject).to include('href="http://www.google.com"')
+        expect(subject).to include('href="http://www.google.com/"')
       end
     end