From 0b0b1dabdfa2051d0b1d1ad846892da4801ca588 Mon Sep 17 00:00:00 2001
From: tusooa <tusooa@kazv.moe>
Date: Fri, 21 Jul 2023 13:54:10 -0400
Subject: Fix parsing non-ascii tags

---
 src/services/matcher/matcher.service.js | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

(limited to 'src')

diff --git a/src/services/matcher/matcher.service.js b/src/services/matcher/matcher.service.js
index b6c4e909..54f02d31 100644
--- a/src/services/matcher/matcher.service.js
+++ b/src/services/matcher/matcher.service.js
@@ -14,8 +14,11 @@ export const mentionMatchesUrl = (attention, url) => {
  * @param {string} url
  */
 export const extractTagFromUrl = (url) => {
-  const regex = /tag[s]*\/(\w+)$/g
-  const result = regex.exec(url)
+  const decoded = decodeURI(url)
+  // https://git.pleroma.social/pleroma/elixir-libraries/linkify/-/blob/master/lib/linkify/parser.ex
+  // https://www.pcre.org/original/doc/html/pcrepattern.html
+  const regex = /tag[s]*\/([\p{L}\p{N}_]*[\p{Alphabetic}_·\u{200c}][\p{L}\p{N}_·\p{M}\u{200c}]*)$/ug
+  const result = regex.exec(decoded)
   if (!result) {
     return false
   }
-- 
cgit v1.2.3-70-g09d2