From 5e7a7701b94ee63927750064a39b188d9e17164a Mon Sep 17 00:00:00 2001
From: Mikael Nordfeldth <mmn@hethane.se>
Date: Sat, 22 Apr 2017 11:07:38 +0200
Subject: [PATCH] Domain name regular expression into lib/framework.php

cherry-pick-merge
---
 lib/framework.php                 | 11 +++++++++++
 plugins/OStatus/OStatusPlugin.php |  4 ++--
 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/lib/framework.php b/lib/framework.php
index 620730370f..29f1d694d4 100644
--- a/lib/framework.php
+++ b/lib/framework.php
@@ -57,6 +57,17 @@ define('NOTICE_INBOX_SOURCE_FORWARD', 4);
 define('NOTICE_INBOX_SOURCE_PROFILE_TAG', 5);
 define('NOTICE_INBOX_SOURCE_GATEWAY', -1);
 
+/**
+ * StatusNet had this string as valid path characters: '\pN\pL\,\!\(\)\.\:\-\_\+\/\=\&\;\%\~\*\$\'\@'
+ * Some of those characters can be troublesome when auto-linking plain text. Such as "http://some.com/)"
+ * URL encoding should be used whenever a weird character is used, the following strings are not definitive.
+ */
+define('URL_REGEX_VALID_PATH_CHARS',        '\pN\pL\,\!\.\:\-\_\+\/\@\=\;\%\~\*');
+define('URL_REGEX_VALID_QSTRING_CHARS',     URL_REGEX_VALID_PATH_CHARS    . '\&');
+define('URL_REGEX_VALID_FRAGMENT_CHARS',    URL_REGEX_VALID_QSTRING_CHARS . '\?\#');
+define('URL_REGEX_EXCLUDED_END_CHARS',      '\?\.\,\!\#\:\'');  // don't include these if they are directly after a URL
+define('URL_REGEX_DOMAIN_NAME', '(?:(?!-)[A-Za-z0-9\-]{1,63}(?<!-)\.)+[A-Za-z]{2,10}');
+
 // append our extlib dir as the last-resort place to find libs
 
 set_include_path(get_include_path() . PATH_SEPARATOR . INSTALLDIR . '/extlib/');
diff --git a/plugins/OStatus/OStatusPlugin.php b/plugins/OStatus/OStatusPlugin.php
index ddee3e9f9c..8a99620aef 100644
--- a/plugins/OStatus/OStatusPlugin.php
+++ b/plugins/OStatus/OStatusPlugin.php
@@ -280,7 +280,7 @@ class OStatusPlugin extends Plugin
     static function extractWebfingerIds($text)
     {
         $wmatches = array();
-        $result = preg_match_all('/(?:^|\s+)@((?:\w+[\w\-\_\.]?)*(?:[\w\-\_\.]*\w+)@(?:(?!-)[A-Za-z0-9\-]{1,63}(?<!-)\.)+[A-Za-z]{2,10})/',
+        $result = preg_match_all('/(?:^|\s+)@((?:\w+[\w\-\_\.]?)*(?:[\w\-\_\.]*\w+)@'.URL_REGEX_DOMAIN_NAME.')/',
                        $text,
                        $wmatches,
                        PREG_OFFSET_CAPTURE);
@@ -300,7 +300,7 @@ class OStatusPlugin extends Plugin
     static function extractUrlMentions($text)
     {
         $wmatches = array();
-        $result = preg_match_all('!(?:^|\s+)@((?:\w+\.)*\w+(?:\w+\-\w+)*\.\w+(?:/\w+)*)!',
+        $result = preg_match_all('!(?:^|\s+)@'.URL_REGEX_DOMAIN_NAME.'(?:/\w+)*)!',
                        $text,
                        $wmatches,
                        PREG_OFFSET_CAPTURE);
-- 
2.39.5