system/core/Utf8.php

   1 <?php  if ( ! defined('BASEPATH')) exit('No direct script access allowed');
   2 /**
   3  * CodeIgniter
   4  *
   5  * An open source application development framework for PHP 5.1.6 or newer
   6  *
   7  * @package             CodeIgniter
   8  * @author              ExpressionEngine Dev Team
   9  * @copyright   Copyright (c) 2008 - 2011, EllisLab, Inc.
  10  * @license             http://codeigniter.com/user_guide/license.html
  11  * @link                http://codeigniter.com
  12  * @since               Version 2.0
  13  * @filesource
  14  */
  15
  16 // ------------------------------------------------------------------------
  17
  18 /**
  19  * Utf8 Class
  20  *
  21  * Provides support for UTF-8 environments
  22  *
  23  * @package             CodeIgniter
  24  * @subpackage  Libraries
  25  * @category    UTF-8
  26  * @author              ExpressionEngine Dev Team
  27  * @link                http://codeigniter.com/user_guide/libraries/utf8.html
  28  */
  29 class CI_Utf8 {
  30
  31         /**
  32          * Constructor
  33          *
  34          * Determines if UTF-8 support is to be enabled
  35          *
  36          */
  37         function __construct()
  38         {
  39                 log_message('debug', "Utf8 Class Initialized");
  40
  41                 global $CFG;
  42
  43                 if (
  44                         preg_match('/./u', 'é') === 1                                  // PCRE must support UTF-8
  45                         AND function_exists('iconv')                                    // iconv must be installed
  46                         AND ini_get('mbstring.func_overload') != 1              // Multibyte string function overloading cannot be enabled
  47                         AND $CFG->item('charset') == 'UTF-8'                    // Application charset must be UTF-8
  48                         )
  49                 {
  50                         log_message('debug', "UTF-8 Support Enabled");
  51
  52                         define('UTF8_ENABLED', TRUE);
  53
  54                         // set internal encoding for multibyte string functions if necessary
  55                         // and set a flag so we don't have to repeatedly use extension_loaded()
  56                         // or function_exists()
  57                         if (extension_loaded('mbstring'))
  58                         {
  59                                 define('MB_ENABLED', TRUE);
  60                                 mb_internal_encoding('UTF-8');
  61                         }
  62                         else
  63                         {
  64                                 define('MB_ENABLED', FALSE);
  65                         }
  66                 }
  67                 else
  68                 {
  69                         log_message('debug', "UTF-8 Support Disabled");
  70                         define('UTF8_ENABLED', FALSE);
  71                 }
  72         }
  73
  74         // --------------------------------------------------------------------
  75
  76         /**
  77          * Clean UTF-8 strings
  78          *
  79          * Ensures strings are UTF-8
  80          *
  81          * @access      public
  82          * @param       string
  83          * @return      string
  84          */
  85         function clean_string($str)
  86         {
  87                 if ($this->_is_ascii($str) === FALSE)
  88                 {
  89                         $str = @iconv('UTF-8', 'UTF-8//IGNORE', $str);
  90                 }
  91
  92                 return $str;
  93         }
  94
  95         // --------------------------------------------------------------------
  96
  97         /**
  98          * Remove ASCII control characters
  99          *
 100          * Removes all ASCII control characters except horizontal tabs,
 101          * line feeds, and carriage returns, as all others can cause
 102          * problems in XML
 103          *
 104          * @access      public
 105          * @param       string
 106          * @return      string
 107          */
 108         function safe_ascii_for_xml($str)
 109         {
 110                 return remove_invisible_characters($str, FALSE);
 111         }
 112
 113         // --------------------------------------------------------------------
 114
 115         /**
 116          * Convert to UTF-8
 117          *
 118          * Attempts to convert a string to UTF-8
 119          *
 120          * @access      public
 121          * @param       string
 122          * @param       string  - input encoding
 123          * @return      string
 124          */
 125         function convert_to_utf8($str, $encoding)
 126         {
 127                 if (function_exists('iconv'))
 128                 {
 129                         $str = @iconv($encoding, 'UTF-8', $str);
 130                 }
 131                 elseif (function_exists('mb_convert_encoding'))
 132                 {
 133                         $str = @mb_convert_encoding($str, 'UTF-8', $encoding);
 134                 }
 135                 else
 136                 {
 137                         return FALSE;
 138                 }
 139
 140                 return $str;
 141         }
 142
 143         // --------------------------------------------------------------------
 144
 145         /**
 146          * Is ASCII?
 147          *
 148          * Tests if a string is standard 7-bit ASCII or not
 149          *
 150          * @access      public
 151          * @param       string
 152          * @return      bool
 153          */
 154         function _is_ascii($str)
 155         {
 156                 return (preg_match('/[^\x00-\x7F]/S', $str) == 0);
 157         }
 158
 159         // --------------------------------------------------------------------
 160
 161 }
 162 // End Utf8 Class
 163
 164 /* End of file Utf8.php */
 165 /* Location: ./system/core/Utf8.php */