From 30ef9fccf0c603ba917191ddbefdd497523d3d67 Mon Sep 17 00:00:00 2001
From: Claire <claire.github-309c@sitedethib.com>
Date: Mon, 6 May 2024 14:47:19 +0200
Subject: [PATCH] Fix hashtag matching pattern matching some link anchors
 (#30190)

---
 app/models/tag.rb       | 2 +-
 spec/models/tag_spec.rb | 4 ++++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/app/models/tag.rb b/app/models/tag.rb
index 35be921e2d..3f88cb0680 100644
--- a/app/models/tag.rb
+++ b/app/models/tag.rb
@@ -37,7 +37,7 @@ class Tag < ApplicationRecord
   HASHTAG_LAST_SEQUENCE = '([[:word:]_]*[[:alpha:]][[:word:]_]*)'
   HASHTAG_NAME_PAT = "#{HASHTAG_FIRST_SEQUENCE}|#{HASHTAG_LAST_SEQUENCE}"
 
-  HASHTAG_RE = %r{(?<![=/)\w])#(#{HASHTAG_NAME_PAT})}i
+  HASHTAG_RE = %r{(?<![=/)\p{Alnum}])#(#{HASHTAG_NAME_PAT})}i
   HASHTAG_NAME_RE = /\A(#{HASHTAG_NAME_PAT})\z/i
   HASHTAG_INVALID_CHARS_RE = /[^[:alnum:]\u0E47-\u0E4E#{HASHTAG_SEPARATORS}]/
 
diff --git a/spec/models/tag_spec.rb b/spec/models/tag_spec.rb
index 4c2bdd52f4..7799afe44b 100644
--- a/spec/models/tag_spec.rb
+++ b/spec/models/tag_spec.rb
@@ -36,6 +36,10 @@ RSpec.describe Tag do
       expect(subject.match('https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111895#c4')).to be_nil
     end
 
+    it 'does not match URLs with hashtag-like anchors after a non-ascii character' do
+      expect(subject.match('https://example.org/testé#foo')).to be_nil
+    end
+
     it 'does not match URLs with hashtag-like anchors after an empty query parameter' do
       expect(subject.match('https://en.wikipedia.org/wiki/Ghostbusters_(song)?foo=#Lawsuit')).to be_nil
     end