utf8/from_unicode.php

   1 <?php defined('SYSPATH') or die('No direct script access.');
   2 /**
   3  * UTF8::from_unicode
   4  *
   5  * @package    Kohana
   6  * @author     Kohana Team
   7  * @copyright  (c) 2007-2011 Kohana Team
   8  * @copyright  (c) 2005 Harry Fuecks
   9  * @license    http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt
  10  */
  11 function _from_unicode($arr)
  12 {
  13         ob_start();
  14
  15         $keys = array_keys($arr);
  16
  17         foreach ($keys as $k)
  18         {
  19                 // ASCII range (including control chars)
  20                 if (($arr[$k] >= 0) AND ($arr[$k] <= 0x007f))
  21                 {
  22                         echo chr($arr[$k]);
  23                 }
  24                 // 2 byte sequence
  25                 elseif ($arr[$k] <= 0x07ff)
  26                 {
  27                         echo chr(0xc0 | ($arr[$k] >> 6));
  28                         echo chr(0x80 | ($arr[$k] & 0x003f));
  29                 }
  30                 // Byte order mark (skip)
  31                 elseif ($arr[$k] == 0xFEFF)
  32                 {
  33                         // nop -- zap the BOM
  34                 }
  35                 // Test for illegal surrogates
  36                 elseif ($arr[$k] >= 0xD800 AND $arr[$k] <= 0xDFFF)
  37                 {
  38                         // Found a surrogate
  39                         throw new UTF8_Exception("UTF8::from_unicode: Illegal surrogate at index: ':index', value: ':value'", array(
  40                                 ':index' => $k,
  41                                 ':value' => $arr[$k],
  42                         ));
  43                 }
  44                 // 3 byte sequence
  45                 elseif ($arr[$k] <= 0xffff)
  46                 {
  47                         echo chr(0xe0 | ($arr[$k] >> 12));
  48                         echo chr(0x80 | (($arr[$k] >> 6) & 0x003f));
  49                         echo chr(0x80 | ($arr[$k] & 0x003f));
  50                 }
  51                 // 4 byte sequence
  52                 elseif ($arr[$k] <= 0x10ffff)
  53                 {
  54                         echo chr(0xf0 | ($arr[$k] >> 18));
  55                         echo chr(0x80 | (($arr[$k] >> 12) & 0x3f));
  56                         echo chr(0x80 | (($arr[$k] >> 6) & 0x3f));
  57                         echo chr(0x80 | ($arr[$k] & 0x3f));
  58                 }
  59                 // Out of range
  60                 else
  61                 {
  62                         throw new UTF8_Exception("UTF8::from_unicode: Codepoint out of Unicode range at index: ':index', value: ':value'", array(
  63                                 ':index' => $k,
  64                                 ':value' => $arr[$k],
  65                         ));
  66                 }
  67         }
  68
  69         $result = ob_get_contents();
  70         ob_end_clean();
  71         return $result;
  72 }