functions/encode/utf_8.php

   1 <?php
   2 /**
   3  * utf-8 encoding function
   4  *
   5  * takes a string of unicode entities and converts it to a utf-8 encoded string
   6  * each unicode entitiy has the form &#nnn(nn); n={0..9} and can be displayed by utf-8 supporting
   7  * browsers.  Ascii will not be modified.
   8  *
   9  * code is taken from www.php.net manual comments
  10  * Author: ronen at greyzone dot com
  11  *
  12  * @version $Id$
  13  * @package squirrelmail
  14  * @subpackage encode
  15  */
  16
  17 /**
  18  * Converts string to utf-8
  19  * @param $source string of unicode entities [STRING]
  20  * @return a utf-8 encoded string [STRING]
  21  */
  22 function charset_encode_utf_8 ($source) {
  23    $utf8Str = '';
  24    $entityArray = explode ("&#", $source);
  25    $size = count ($entityArray);
  26    for ($i = 0; $i < $size; $i++) {
  27        $subStr = $entityArray[$i];
  28        $nonEntity = strstr ($subStr, ';');
  29        if ($nonEntity !== false) {
  30            $unicode = intval (substr ($subStr, 0, (strpos ($subStr, ';') + 1)));
  31            // determine how many chars are needed to reprsent this unicode char
  32            if ($unicode < 128) {
  33                $utf8Substring = chr ($unicode);
  34            }
  35            else if ($unicode >= 128 && $unicode < 2048) {
  36                $binVal = str_pad (decbin ($unicode), 11, "0", STR_PAD_LEFT);
  37                $binPart1 = substr ($binVal, 0, 5);
  38                $binPart2 = substr ($binVal, 5);
  39
  40                $char1 = chr (192 + bindec ($binPart1));
  41                $char2 = chr (128 + bindec ($binPart2));
  42                $utf8Substring = $char1 . $char2;
  43            }
  44            else if ($unicode >= 2048 && $unicode < 65536) {
  45                $binVal = str_pad (decbin ($unicode), 16, "0", STR_PAD_LEFT);
  46                $binPart1 = substr ($binVal, 0, 4);
  47                $binPart2 = substr ($binVal, 4, 6);
  48                $binPart3 = substr ($binVal, 10);
  49
  50                $char1 = chr (224 + bindec ($binPart1));
  51                $char2 = chr (128 + bindec ($binPart2));
  52                $char3 = chr (128 + bindec ($binPart3));
  53                $utf8Substring = $char1 . $char2 . $char3;
  54            }
  55            else {
  56                $binVal = str_pad (decbin ($unicode), 21, "0", STR_PAD_LEFT);
  57                $binPart1 = substr ($binVal, 0, 3);
  58                $binPart2 = substr ($binVal, 3, 6);
  59                $binPart3 = substr ($binVal, 9, 6);
  60                $binPart4 = substr ($binVal, 15);
  61
  62                $char1 = chr (240 + bindec ($binPart1));
  63                $char2 = chr (128 + bindec ($binPart2));
  64                $char3 = chr (128 + bindec ($binPart3));
  65                $char4 = chr (128 + bindec ($binPart4));
  66                $utf8Substring = $char1 . $char2 . $char3 . $char4;
  67            }
  68
  69            if (strlen ($nonEntity) > 1)
  70                $nonEntity = substr ($nonEntity, 1); // chop the first char (';')
  71            else
  72                $nonEntity = '';
  73
  74            $utf8Str .= $utf8Substring . $nonEntity;
  75        }
  76        else {
  77            $utf8Str .= $subStr;
  78        }
  79    }
  80
  81    return $utf8Str;
  82 }
  83 ?>