Switched to full UTF-8 support

3 files changed, 114 insertions, 83 deletions

diff --git a/program/include/main.inc b/program/include/main.inc
index c64ac7d38..e40ab0948 100644
--- a/program/include/main.inc
+++ b/program/include/main.inc
@@ -20,6 +20,7 @@
 */
 
 require_once('lib/des.inc');
+require_once('lib/utf7.inc');
 
 
 // register session and connect to server
@@ -81,7 +82,7 @@ function rcmail_startup($task='mail')
   if (!$_SESSION['client_id'])
     {
     $_SESSION['client_id'] = $sess_id;
-    $_SESSION['user_lang'] = substr($CONFIG['locale_string'], 0, 2);
+    $_SESSION['user_lang'] = rcube_language_prop($CONFIG['locale_string']);
     $_SESSION['auth_time'] = mktime();
     $_SESSION['auth'] = rcmail_auth_hash($sess_id, $_SESSION['auth_time']);
     unset($GLOBALS['_auth']);
@@ -89,7 +90,7 @@ function rcmail_startup($task='mail')
 
   // set session vars global
   $sess_auth = $_SESSION['auth'];
-  $sess_user_lang = $_SESSION['user_lang'];
+  $sess_user_lang = rcube_language_prop($_SESSION['user_lang']);
 
 
   // overwrite config with user preferences
@@ -200,6 +201,9 @@ function rcmail_shutdown()
     $IMAP->close();
     $IMAP->write_cache();
     }
+    
+  // before closing the database connection, write session data
+  session_write_close();
   }
 
 
@@ -244,7 +248,42 @@ function get_sequence_name($sequence)
   
   return $table;
   }
-  
+
+
+// check the given string and returns language properties
+function rcube_language_prop($lang, $prop='lang')
+  {
+  global $INSTLL_PATH;
+  static $rcube_languages, $rcube_language_aliases, $rcube_charsets;
+
+  if (empty($rcube_languages))
+    @include($INSTLL_PATH.'program/localization/index.inc');
+    
+  // check if we have an alias for that language
+  if (!isset($rcube_languages[$lang]) && isset($rcube_language_aliases[$lang]))
+    $lang = $rcube_language_aliases[$lang];
+    
+  // try the first two chars
+  if (!isset($rcube_languages[$lang]) && strlen($lang>2))
+    {
+    $lang = substr($lang, 0, 2);
+    $lang = rcube_language_prop($lang);
+    }
+
+  if (!isset($rcube_languages[$lang]))
+    $lang = 'en_US';
+
+  // language has special charset configured
+  if (isset($rcube_charsets[$lang]))
+    $charset = $rcube_charsets[$lang];
+  else
+    $charset = 'UTF-8';    
+
+  if ($prop=='charset')
+    return $charset;
+  else
+    return $lang;
+  }
   
 
 // init output object for GUI and add common scripts
@@ -273,32 +312,22 @@ function load_gui()
     $CHARSET = $CONFIG['charset'];
     }
   else
+    {
     rcmail_set_locale($sess_user_lang);
-  
+    }
+
   // add some basic label to client
   rcube_add_label('loading');
-  }  
+  }
 
 
 // set localization charset based on the given language
 function rcmail_set_locale($lang)
   {
-  global $OUTPUT, $INSTLL_PATH, $CHARSET;
-  static $rcube_charsets;
-
-  if (!$rcube_charsets)
-    @include($INSTLL_PATH.'program/localization/index.inc');
+  global $OUTPUT, $CHARSET;
 
-  if (isset($rcube_charsets[$lang]))
-    {
-    $OUTPUT->set_charset($rcube_charsets[$lang]);
-    $CHARSET = $rcube_charsets[$lang];
-    }
-  else
-    {
-    $OUTPUT->set_charset('ISO-8859-1');
-    $CHARSET = 'ISO-8859-1';
-    }
+  $CHARSET = rcube_language_prop($lang, 'charset');
+  $OUTPUT->set_charset($CHARSET);
   }
 
 
@@ -675,6 +704,35 @@ function rcmail_clear_session_temp($sess_id)
 
 
 
+// decode characters from any charset to UTF-8
+function decode_specialchars($input, $charset='')
+  {
+  global $CHARSET;
+  $charset = strtolower($charset);
+  
+  if ($charset=='utf-7')
+    $output = UTF7DecodeString($input);
+  else if ($charset=='utf-8' && $CHARSET!='UTF-8' && function_exists('utf8_decode'))
+    $output = utf8_decode($input);
+  else if ($charset=="koi8-r")
+    $output = convert_cyr_string($input, 'k', 'w');
+  else if ($charset=="iso8859-5")
+    $output = convert_cyr_string($input, 'i', 'w');
+  else if ($charset=="x-cp866")
+    $output = convert_cyr_string($input, 'a', 'w');
+  else if ($charset=="x-mac-cyrillic")
+    $output = convert_cyr_string($input, 'm', 'w');
+  else
+    $output = $input;
+    
+  // encode string for output
+  if ($CHARSET=='UTF-8' && $charset!='utf-8' && function_exists('utf8_encode'))
+    $output = utf8_encode($output);
+
+  return $output;
+  }
+
+
 // replace specials characters to a specific encoding type
 function rep_specialchars_output($str, $enctype='', $mode='', $newlines=TRUE)
   {
@@ -688,7 +746,6 @@ function rep_specialchars_output($str, $enctype='', $mode='', $newlines=TRUE)
   if ($enctype!='html')
     $str = str_replace(chr(160), ' ', $str);
 
-
   // encode for plaintext
   if ($enctype=='text')
     return str_replace("\r\n", "\n", $mode=='remove' ? strip_tags($str) : $str);
@@ -698,14 +755,7 @@ function rep_specialchars_output($str, $enctype='', $mode='', $newlines=TRUE)
     {
     if (!$html_encode_arr)
       {
-      if ($CHARSET=='ISO-8859-1')
-        {
-        $html_encode_arr = get_html_translation_table(HTML_ENTITIES);
-        $html_encode_arr[chr(128)] = '€';
-        }
-      else
-        $html_encode_arr = get_html_translation_table(HTML_SPECIALCHARS);
-        
+      $html_encode_arr = get_html_translation_table(HTML_SPECIALCHARS);        
       unset($html_encode_arr['?']);
       unset($html_encode_arr['&']);
       }
@@ -724,7 +774,7 @@ function rep_specialchars_output($str, $enctype='', $mode='', $newlines=TRUE)
       $str = strip_tags($str);
       
     $out = strtr($str, $encode_arr);
-
+      
     return $newlines ? nl2br($out) : $out;
     }
 
@@ -769,6 +819,26 @@ function rep_specialchars_output($str, $enctype='', $mode='', $newlines=TRUE)
   }
 
 
+// convert a string from one charset to another
+// this function is not complete and not tested well
+function rcube_charset_convert($str, $from, $to)
+  {
+  $charset = strtolower($to);
+  
+  if (strtolower($from)==$charset)
+    return $str;
+  
+  $str = decode_specialchars($str, $from);
+  
+  if ($charset=='utf-8' && function_exists('utf8_encode'))
+    return utf8_encode($str);
+  else if ($charset=='utf-7')
+    return UTF7EncodeString($str);
+
+  return $str;
+  }
+  
+
 
 // ************** template parsing and gui functions **************
 
@@ -841,7 +911,7 @@ function parse_rcube_xml($input)
 
 function rcube_xml_command($command, $str_attrib, $a_attrib=NULL)
   {
-  global $IMAP, $CONFIG;
+  global $IMAP, $CONFIG, $OUTPUT;
   
   $attrib = array();
   $command = strtolower($command);
@@ -952,7 +1022,7 @@ function rcube_xml_command($command, $str_attrib, $a_attrib=NULL)
         else if (isset($GLOBALS['PAGE_TITLE']))
           $title .= $GLOBALS['PAGE_TITLE'];
         else if ($task=='mail' && ($mbox_name = $IMAP->get_mailbox_name()))
-          $title .= UTF7DecodeString($mbox_name);
+          $title .= $OUTPUT->encode_string(UTF7DecodeString($mbox_name));
         else
           $title .= $task;
           
diff --git a/program/include/rcube_db.inc b/program/include/rcube_db.inc
index e54dcc989..b935e1d15 100755
--- a/program/include/rcube_db.inc
+++ b/program/include/rcube_db.inc
@@ -81,16 +81,6 @@ class rcube_db
 
 
   /**
-   * Object destructor
-   */
-  function __destruct()
-    {
-    // before closing the database connection, write session data
-    session_write_close();
-    } 
-
-
-  /**
    * Connect to specific database
    *
    * @param  string  DSN for DB connections
diff --git a/program/include/rcube_shared.inc b/program/include/rcube_shared.inc
index 75db7603f..b622203dd 100644
--- a/program/include/rcube_shared.inc
+++ b/program/include/rcube_shared.inc
@@ -87,6 +87,15 @@ class rcube_html_page
     {
     $this->charset = $charset;
     }
+    
+    
+  function encode_string($str)
+    {
+    if ($this->charset=='UTF-8' && function_exists('utf8_encode'))
+      return utf8_encode($str);
+    else
+      return $str;
+    }
 
 
   function write($templ='', $base_path='')
@@ -1062,8 +1071,8 @@ function rcube_label($attrib)
     $sa_text_data = array();
     
     // get english labels (these should be complete)
-    @include($INSTALL_PATH.'program/localization/en/labels.inc');
-    @include($INSTALL_PATH.'program/localization/en/messages.inc');
+    @include($INSTALL_PATH.'program/localization/en_US/labels.inc');
+    @include($INSTALL_PATH.'program/localization/en_US/messages.inc');
 
     if (is_array($labels))
       $sa_text_data = $labels;
@@ -1075,19 +1084,13 @@ function rcube_label($attrib)
       {
       include_once($INSTALL_PATH.'program/localization/'.$sess_user_lang.'/labels.inc');
       include_once($INSTALL_PATH.'program/localization/'.$sess_user_lang.'/messages.inc');
-      
+
       if (is_array($labels))
         $sa_text_data = array_merge($sa_text_data, $labels);
       if (is_array($messages))
         $sa_text_data = array_merge($sa_text_data, $messages);
       }
       
-    if (isset($utf8_decoding) && $utf8_decoding==TRUE)
-      {
-      @include_once('lib/utf8.inc');
-      $utf8_decode = TRUE;
-      }
-      
     $s_language = $sess_user_lang;
     }
 
@@ -1142,14 +1145,6 @@ $text
 EOF;
 ");
 
-
-  // perform utf-8 decoding
-  if ($utf8_decode && function_exists('utf8ToUnicodeEntities'))
-    $text = utf8ToUnicodeEntities($text);
-  else if ($utf8_decode)
-    $OUTPUT->set_charset('UTF-8');
-
-
   // format output
   if (($attrib['uppercase'] && strtolower($attrib['uppercase']=='first')) || $attrib['ucfirst'])
     return ucfirst($text);
@@ -1185,30 +1180,6 @@ function send_future_expire_header()
   }
 
 
-
-function decode_specialchars($input, $charset='')
-  {
-  $charset = strtolower($charset);
-  
-  if ($charset=='utf-8')
-    {
-    require_once('lib/utf8.inc');
-    return utf8ToUnicodeEntities($input);
-    }
-  else if ($charset=="koi8-r")
-    return convert_cyr_string($input, 'k', 'w');
-  else if ($charset=="iso8859-5")
-    return convert_cyr_string($input, 'i', 'w');
-  else if ($charset=="x-cp866")
-    return convert_cyr_string($input, 'a', 'w');
-  else if ($charset=="x-mac-cyrillic")
-    return convert_cyr_string($input, 'm', 'w');
-
-  return $input;
-  }
-
-
-
 // function to convert an array to a javascript array
 function array2js($arr, $type='')
   {