Fix URL linkifier grabbing full-width spaces and quotations (mastodon…

…#9997)

Fix mastodon#9993
Fix mastodon#5654

app/lib/formatter.rb

@@ -204,12 +204,22 @@ def rewrite(text, entities)
     result.flatten.join
   end
 
+  UNICODE_ESCAPE_BLACKLIST_RE = /\p{Z}|\p{P}/
+
   def utf8_friendly_extractor(text, options = {})
     old_to_new_index = [0]
 
     escaped = text.chars.map do |c|
-      output = c.ord.to_s(16).length > 2 ? CGI.escape(c) : c
+      output = begin
+        if c.ord.to_s(16).length > 2 && UNICODE_ESCAPE_BLACKLIST_RE.match(c).nil?
+          CGI.escape(c)
+        else
+          c
+        end
+      end
+
       old_to_new_index << old_to_new_index.last + output.length
+
       output
     end.join
 

config/initializers/twitter_regex.rb

@@ -1,7 +1,7 @@
 module Twitter
   class Regex
-    REGEXEN[:valid_general_url_path_chars] = /[^\p{White_Space}\(\)\?]/iou
-    REGEXEN[:valid_url_path_ending_chars] = /[^\p{White_Space}\(\)\?!\*';:=\,\.\$%\[\]~&\|@]|(?:#{REGEXEN[:valid_url_balanced_parens]})/iou
+    REGEXEN[:valid_general_url_path_chars] = /[^\p{White_Space}<>\(\)\?]/iou
+    REGEXEN[:valid_url_path_ending_chars] = /[^\p{White_Space}\(\)\?!\*"'「」<>;:=\,\.\$%\[\]~&\|@]|(?:#{REGEXEN[:valid_url_balanced_parens]})/iou
     REGEXEN[:valid_url_balanced_parens] = /
       \(
         (?:

spec/lib/formatter_spec.rb

@@ -115,6 +115,22 @@
       end
     end
 
+    context 'given a URL in quotation marks' do
+      let(:text) { '"https://example.com/"' }
+
+      it 'does not match the quotation marks' do
+        is_expected.to include 'href="https://example.com/"'
+      end
+    end
+
+    context 'given a URL in angle brackets' do
+      let(:text) { '<https://example.com/>' }
+
+      it 'does not match the angle brackets' do
+        is_expected.to include 'href="https://example.com/"'
+      end
+    end
+
     context 'given a URL with Japanese path string' do
       let(:text) { 'https://ja.wikipedia.org/wiki/日本' }
 
@@ -131,6 +147,22 @@
       end
     end
 
+    context 'given a URL with a full-width space' do
+      let(:text) { 'https://example.com/　abc123' }
+
+      it 'does not match the full-width space' do
+        is_expected.to include 'href="https://example.com/"'
+      end
+    end
+
+    context 'given a URL in Japanese quotation marks' do
+      let(:text) { '「[https://example.org/」' }
+
+      it 'does not match the quotation marks' do
+        is_expected.to include 'href="https://example.org/"'
+      end
+    end
+
     context 'given a URL with Simplified Chinese path string' do
       let(:text) { 'https://baike.baidu.com/item/中华人民共和国' }
 
@@ -150,15 +182,23 @@
     context 'given a URL containing unsafe code (XSS attack, visible part)' do
       let(:text) { %q{http://example.com/b<del>b</del>} }
 
-      it 'escapes the HTML in the URL' do
+      it 'does not include the HTML in the URL' do
+        is_expected.to include '"http://example.com/b"'
+      end
+
+      it 'escapes the HTML' do
         is_expected.to include '&lt;del&gt;b&lt;/del&gt;'
       end
     end
 
     context 'given a URL containing unsafe code (XSS attack, invisible part)' do
       let(:text) { %q{http://example.com/blahblahblahblah/a<script>alert("Hello")</script>} }
 
-      it 'escapes the HTML in the URL' do
+      it 'does not include the HTML in the URL' do
+        is_expected.to include '"http://example.com/blahblahblahblah/a"'
+      end
+
+      it 'escapes the HTML' do
         is_expected.to include '&lt;script&gt;alert(&quot;Hello&quot;)&lt;/script&gt;'
       end
     end