Skip iconv for problematic ISO-2022-JP strings (#5668)

We sometimes get broken character encodings such as:
Subject: =?iso-2022-jp?B?GyRCLWo7M3l1OSk2SBsoQgo=?=
This actually is not a strict ISO-2022-JP string, but a CP50220 string
that is a variant of ISO-2022-JP with extended characters proposed by
Microsoft. Iconv can not handle these encodings well.

program/lib/Roundcube/rcube_charset.php

@@ -201,7 +201,9 @@ class rcube_charset
         }
 
         // convert charset using iconv module
-        if ($iconv_options !== false && $from != 'UTF7-IMAP' && $to != 'UTF7-IMAP') {
+        if ($iconv_options !== false && $from != 'UTF7-IMAP' && $to != 'UTF7-IMAP'
+            && $from !== 'ISO-2022-JP'
+        ) {
             // throw an exception if iconv reports an illegal character in input
             // it means that input string has been truncated
             set_error_handler(array('rcube_charset', 'error_handler'), E_NOTICE);
@@ -227,6 +229,7 @@ class rcube_charset
             $aliases = array(
                 'WINDOWS-1257' => 'ISO-8859-13',
                 'US-ASCII'     => 'ASCII',
+                'ISO-2022-JP'  => 'ISO-2022-JP-MS',
             );
 
             $mb_from = $aliases[$from] ?: $from;

tests/Framework/Charset.php

@@ -70,6 +70,7 @@ class Framework_Charset extends PHPUnit_Framework_TestCase
             array('aż', 'a', 'UTF-8', 'US-ASCII'),
             array('&BCAEMARBBEEESwQ7BDoEOA-', 'Рассылки', 'UTF7-IMAP', 'UTF-8'),
             array('Рассылки', '&BCAEMARBBEEESwQ7BDoEOA-', 'UTF-8', 'UTF7-IMAP'),
+            array(base64_decode('GyRCLWo7M3l1OSk2SBsoQg=='), '㈱山﨑工業', 'ISO-2022-JP', 'UTF-8'),
         );
     }
 
@@ -178,7 +179,7 @@ class Framework_Charset extends PHPUnit_Framework_TestCase
     function data_detect_with_lang()
     {
         return array(
-            array('顯示名稱,主要', 'zh_TW', 'BIG-5'),
+            array(base64_decode('xeOl3KZXutkspUStbg=='), 'zh_TW', 'BIG-5'),
         );
     }