- Fix handling of top-level domains with more than 5 chars or unicode chars (#1487883)

CHANGELOG

@@ -1,6 +1,7 @@
 CHANGELOG Roundcube Webmail
 ===========================
 
+- Fix handling of top-level domains with more than 5 chars or unicode chars (#1487883)
 - Fix usage of non-standard HTTP error codes (#1487797)
 - Improve performance by including files with absolute path (#1487849)
 - Move folder name truncation to client/skin (#1485412)

program/include/rcube_string_replacer.php

@@ -36,7 +36,8 @@ class rcube_string_replacer
   function __construct()
   {
     // Simplified domain expression for UTF8 characters handling
-    $utf_domain = '[^?&@"\'\\/()\s\r\t\n]+\\.[a-z]{2,5}';
+    // Support unicode/punycode in top-level domain part
+    $utf_domain = '[^?&@"\'\\/()\s\r\t\n]+\\.([^\\x00-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-z0-9]{2,})';
     $url1 = '.:;';
     $url2 = 'a-z0-9%=#@+?&\\/_~\\[\\]-';
 

program/js/common.js

@@ -493,8 +493,8 @@ function rcube_check_email(input, inline)
       // So, e-mail address should be validated also on server side after idn_to_ascii() use
       //domain_literal = '\\x5b('+dtext+'|'+quoted_pair+')*\\x5d',
       //sub_domain = '('+atom+'|'+domain_literal+')',
-      // allow punycode in last domain part for ICANN test domains
-      domain = '([^@\\x2e]+\\x2e)+([a-z]{2,}|xn--[a-z0-9]{2,})',
+      // allow punycode/unicode top-level domain
+      domain = '([^@\\x2e]+\\x2e)+([^\\x00-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-z0-9]{2,})',
       // ICANN e-mail test (http://idn.icann.org/E-mail_test)
       icann_domains = [
         '\\u0645\\u062b\\u0627\\u0644\\x2e\\u0625\\u062e\\u062a\\u0628\\u0627\\u0631',

program/steps/mail/func.inc

@@ -22,11 +22,11 @@
 // setup some global vars used by mail steps
 $SENT_MBOX = $RCMAIL->config->get('sent_mbox');
 $DRAFTS_MBOX = $RCMAIL->config->get('drafts_mbox');
-$SEARCH_MODS_DEFAULT = array('*' => array('subject'=>1, 'from'=>1), $SENT_MBOX => array('subject'=>1, 'to'=>1), $DRAFTS_MBOX => array('subject'=>1, 'to'=>1));
-
-// Simplified for IDN in Unicode
-//$EMAIL_ADDRESS_PATTERN = '([a-z0-9][a-z0-9\-\.\+\_]*@[a-z0-9][a-z0-9\-\.]*\\.[a-z]{2,5})';
-$EMAIL_ADDRESS_PATTERN = '([a-z0-9][a-z0-9\-\.\+\_]*@[^&@"\'.][^@&"\']*\\.[a-z]{2,5})';
+$SEARCH_MODS_DEFAULT = array(
+    '*'         => array('subject'=>1, 'from'=>1),
+    $SENT_MBOX  => array('subject'=>1, 'to'=>1),
+    $DRAFTS_MBOX => array('subject'=>1, 'to'=>1)
+);
 
 // actions that do not require imap connection here
 $NOIMAP_ACTIONS = array('addcontact', 'autocomplete', 'upload', 'display-attachment', 'remove-attachment', 'get');
@@ -1197,7 +1197,10 @@ function rcmail_html4inline($body, $container_id, $body_id='', &$attributes=null
  */
 function rcmail_alter_html_link($matches)
 {
-  global $RCMAIL, $EMAIL_ADDRESS_PATTERN;
+  global $RCMAIL;
+
+  // Support unicode/punycode in top-level domain part
+  $EMAIL_PATTERN = '([a-z0-9][a-z0-9\-\.\+\_]*@[^&@"\'.][^@&"\']*\\.([^\\x00-\\x40\\x5b-\\x60\\x7b-\\x7f]{2,}|xn--[a-z0-9]{2,}))';
 
   $tag = $matches[1];
   $attrib = parse_attrib_string($matches[2]);
@@ -1212,12 +1215,12 @@ function rcmail_alter_html_link($matches)
     $attrib['href'] = $RCMAIL->url(array('task' => 'utils', 'action' => 'modcss', 'u' => $tempurl, 'c' => $GLOBALS['rcmail_html_container_id']));
     $end = ' />';
   }
-  else if (preg_match('/^mailto:'.$EMAIL_ADDRESS_PATTERN.'(\?[^"\'>]+)?/i', $attrib['href'], $mailto)) {
+  else if (preg_match('/^mailto:'.$EMAIL_PATTERN.'(\?[^"\'>]+)?/i', $attrib['href'], $mailto)) {
     $attrib['href'] = $mailto[0];
     $attrib['onclick'] = sprintf(
       "return %s.command('compose','%s',this)",
       JS_OBJECT_NAME,
-      JQ($mailto[1].$mailto[2]));
+      JQ($mailto[1].$mailto[3]));
   }
   else if (!empty($attrib['href']) && $attrib['href'][0] != '#') {
     $attrib['target'] = '_blank';