banana/utf8.php

   1 <?php
   2
   3 /**
   4  * Transforme une chaine encodée en UTF-8, et la convertit
   5  * en entitiées unicode &#xxx; pour que ça s'affiche correctement
   6  * dans les navigateurs, sans forcément tenir compte du meta
   7  * content-type charset...
   8  * @param String $source la chaine en UTF-8
   9  * @return String les entitées
  10  * @access public
  11  * @see http://www.php.net/utf8_decode
  12  */
  13 function utf8entities($source)
  14 {
  15    // array used to figure what number to decrement from character order value
  16    // according to number of characters used to map unicode to ascii by utf-8
  17    $decrement[4] = 240;
  18    $decrement[3] = 224;
  19    $decrement[2] = 192;
  20    $decrement[1] = 0;
  21
  22    // the number of bits to shift each charNum by
  23    $shift[1][0] = 0;
  24    $shift[2][0] = 6;
  25    $shift[2][1] = 0;
  26    $shift[3][0] = 12;
  27    $shift[3][1] = 6;
  28    $shift[3][2] = 0;
  29    $shift[4][0] = 18;
  30    $shift[4][1] = 12;
  31    $shift[4][2] = 6;
  32    $shift[4][3] = 0;
  33
  34    $pos = 0;
  35    $len = strlen($source);
  36    $encodedString = '';
  37    while ($pos < $len)
  38    {
  39       $charPos = $source{$pos};
  40       $asciiPos = ord($charPos);
  41       if ($asciiPos < 128)
  42       {
  43          $encodedString .= $charPos;
  44          $pos++;
  45          continue;
  46       }
  47
  48       $i=1;
  49       if (($asciiPos >= 240) && ($asciiPos <= 255)) // 4 chars representing one unicode character
  50          $i=4;
  51       else if (($asciiPos >= 224) && ($asciiPos <= 239)) // 3 chars representing one unicode character
  52          $i=3;
  53       else if (($asciiPos >= 192) && ($asciiPos <= 223)) // 2 chars representing one unicode character
  54          $i=2;
  55       else // 1 char (lower ascii)
  56          $i=1;
  57       $thisLetter = substr($source, $pos, $i);
  58       $pos += $i;
  59
  60       // process the string representing the letter to a unicode entity
  61       $thisLen = strlen($thisLetter);
  62       $thisPos = 0;
  63       $decimalCode = 0;
  64       while ($thisPos < $thisLen)
  65       {
  66          $thisCharOrd = ord(substr($thisLetter, $thisPos, 1));
  67          if ($thisPos == 0)
  68          {
  69             $charNum = intval($thisCharOrd - $decrement[$thisLen]);
  70             $decimalCode += ($charNum << $shift[$thisLen][$thisPos]);
  71          }
  72          else
  73          {
  74             $charNum = intval($thisCharOrd - 128);
  75             $decimalCode += ($charNum << $shift[$thisLen][$thisPos]);
  76          }
  77
  78          $thisPos++;
  79       }
  80
  81       $encodedLetter = '&#'. str_pad($decimalCode, ($thisLen==1)?3:5, '0', STR_PAD_LEFT).';';
  82       $encodedString .= $encodedLetter;
  83    }
  84
  85    return $encodedString;
  86 }
  87
  88 ?>