¤ thing + #294

[banana.git] / banana / utf8.php

<?php\r
\r
/**\r
 * Transforme une chaine encodée en UTF-8, et la convertit\r
 * en entitiées unicode &#xxx; pour que ça s'affiche correctement\r
 * dans les navigateurs, sans forcément tenir compte du meta\r
 * content-type charset...\r
 * @param String $source la chaine en UTF-8\r
 * @return String les entitées\r
 * @access public\r
 * @see http://www.php.net/utf8_decode\r
 */\r
function utf8entities($source)\r
{\r
   // array used to figure what number to decrement from character order value \r
   // according to number of characters used to map unicode to ascii by utf-8\r
   $decrement[4] = 240;\r
   $decrement[3] = 224;\r
   $decrement[2] = 192;\r
   $decrement[1] = 0;\r
   \r
   // the number of bits to shift each charNum by\r
   $shift[1][0] = 0;\r
   $shift[2][0] = 6;\r
   $shift[2][1] = 0;\r
   $shift[3][0] = 12;\r
   $shift[3][1] = 6;\r
   $shift[3][2] = 0;\r
   $shift[4][0] = 18;\r
   $shift[4][1] = 12;\r
   $shift[4][2] = 6;\r
   $shift[4][3] = 0;\r
   \r
   $pos = 0;\r
   $len = strlen($source);\r
   $encodedString = '';\r
   while ($pos < $len)\r
   {\r
      $charPos = $source{$pos};\r
      $asciiPos = ord($charPos);\r
      if ($asciiPos < 128)\r
      {\r
         $encodedString .= $charPos;\r
         $pos++;\r
         continue;\r
      }\r
      \r
      $i=1;\r
      if (($asciiPos >= 240) && ($asciiPos <= 255)) // 4 chars representing one unicode character\r
         $i=4;\r
      else if (($asciiPos >= 224) && ($asciiPos <= 239)) // 3 chars representing one unicode character\r
         $i=3;\r
      else if (($asciiPos >= 192) && ($asciiPos <= 223)) // 2 chars representing one unicode character\r
         $i=2;\r
      else // 1 char (lower ascii)\r
         $i=1;\r
      $thisLetter = substr($source, $pos, $i);\r
      $pos += $i;\r
      \r
      // process the string representing the letter to a unicode entity\r
      $thisLen = strlen($thisLetter);\r
      $thisPos = 0;\r
      $decimalCode = 0;\r
      while ($thisPos < $thisLen)\r
      {\r
         $thisCharOrd = ord(substr($thisLetter, $thisPos, 1));\r
         if ($thisPos == 0)\r
         {\r
            $charNum = intval($thisCharOrd - $decrement[$thisLen]);\r
            $decimalCode += ($charNum << $shift[$thisLen][$thisPos]);\r
         }\r
         else\r
         {\r
            $charNum = intval($thisCharOrd - 128);\r
            $decimalCode += ($charNum << $shift[$thisLen][$thisPos]);\r
         }\r
         \r
         $thisPos++;\r
      }\r
      \r
      $encodedLetter = '&#'. str_pad($decimalCode, ($thisLen==1)?3:5, '0', STR_PAD_LEFT).';';\r
      $encodedString .= $encodedLetter;\r
   }\r
   \r
   return $encodedString;\r
}\r
\r
?>\r