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  * @package squirrelmail
  13  * @subpackage encode
  14  * @param $source string of unicode entities [STRING]
  15  * @return a utf-8 encoded string [STRING]
  16  * @access public
  17  */
  18 function charset_encode_utf_8 ($source) {
  19    $utf8Str = '';
  20    $entityArray = explode ("&#", $source);
  21    $size = count ($entityArray);
  22    for ($i = 0; $i < $size; $i++) {
  23        $subStr = $entityArray[$i];
  24        $nonEntity = strstr ($subStr, ';');
  25        if ($nonEntity !== false) {
  26            $unicode = intval (substr ($subStr, 0, (strpos ($subStr, ';') + 1)));
  27            // determine how many chars are needed to reprsent this unicode char
  28            if ($unicode < 128) {
  29                $utf8Substring = chr ($unicode);
  30            }
  31            else if ($unicode >= 128 && $unicode < 2048) {
  32                $binVal = str_pad (decbin ($unicode), 11, "0", STR_PAD_LEFT);
  33                $binPart1 = substr ($binVal, 0, 5);
  34                $binPart2 = substr ($binVal, 5);
  35
  36                $char1 = chr (192 + bindec ($binPart1));
  37                $char2 = chr (128 + bindec ($binPart2));
  38                $utf8Substring = $char1 . $char2;
  39            }
  40            else if ($unicode >= 2048 && $unicode < 65536) {
  41                $binVal = str_pad (decbin ($unicode), 16, "0", STR_PAD_LEFT);
  42                $binPart1 = substr ($binVal, 0, 4);
  43                $binPart2 = substr ($binVal, 4, 6);
  44                $binPart3 = substr ($binVal, 10);
  45
  46                $char1 = chr (224 + bindec ($binPart1));
  47                $char2 = chr (128 + bindec ($binPart2));
  48                $char3 = chr (128 + bindec ($binPart3));
  49                $utf8Substring = $char1 . $char2 . $char3;
  50            }
  51            else {
  52                $binVal = str_pad (decbin ($unicode), 21, "0", STR_PAD_LEFT);
  53                $binPart1 = substr ($binVal, 0, 3);
  54                $binPart2 = substr ($binVal, 3, 6);
  55                $binPart3 = substr ($binVal, 9, 6);
  56                $binPart4 = substr ($binVal, 15);
  57
  58                $char1 = chr (240 + bindec ($binPart1));
  59                $char2 = chr (128 + bindec ($binPart2));
  60                $char3 = chr (128 + bindec ($binPart3));
  61                $char4 = chr (128 + bindec ($binPart4));
  62                $utf8Substring = $char1 . $char2 . $char3 . $char4;
  63            }
  64
  65            if (strlen ($nonEntity) > 1)
  66                $nonEntity = substr ($nonEntity, 1); // chop the first char (';')
  67            else
  68                $nonEntity = '';
  69
  70            $utf8Str .= $utf8Substring . $nonEntity;
  71        }
  72        else {
  73            $utf8Str .= $subStr;
  74        }
  75    }
  76
  77    return $utf8Str;
  78 }
  79 ?>