banana/text.func.inc.php

   1 <?php
   2 /********************************************************************************
   3 * banana/text.php : text tools
   4 * ---------------
   5 *
   6 * This file is part of the banana distribution
   7 * Copyright: See COPYING files that comes with this distribution
   8 ********************************************************************************/
   9
  10 if (function_exists('dgettext')) {
  11     function _b_($str)
  12     {
  13         return dgettext('banana', $str);
  14     }
  15 } else {
  16     function _b_($str)
  17     {
  18         return $str;
  19     }
  20 }
  21
  22 if (!function_exists('is_utf8')) {
  23     function is_utf8($s)
  24     {
  25         return @iconv('utf-8', 'utf-8', $s) == $s;
  26     }
  27 }
  28
  29 function banana_utf8entities($source)
  30 {
  31    // array used to figure what number to decrement from character order value
  32    // according to number of characters used to map unicode to ascii by utf-8
  33    $decrement[4] = 240;
  34    $decrement[3] = 224;
  35    $decrement[2] = 192;
  36    $decrement[1] = 0;
  37
  38    // the number of bits to shift each charNum by
  39    $shift[1][0] = 0;
  40    $shift[2][0] = 6;
  41    $shift[2][1] = 0;
  42    $shift[3][0] = 12;
  43    $shift[3][1] = 6;
  44    $shift[3][2] = 0;
  45    $shift[4][0] = 18;
  46    $shift[4][1] = 12;
  47    $shift[4][2] = 6;
  48    $shift[4][3] = 0;
  49
  50    $pos = 0;
  51    $len = strlen($source);
  52    $encodedString = '';
  53    while ($pos < $len)
  54    {
  55       $charPos = $source{$pos};
  56       $asciiPos = ord($charPos);
  57       if ($asciiPos < 128)
  58       {
  59          $encodedString .= $charPos;
  60          $pos++;
  61          continue;
  62       }
  63
  64       $i=1;
  65       if (($asciiPos >= 240) && ($asciiPos <= 255)) // 4 chars representing one unicode character
  66          $i=4;
  67       else if (($asciiPos >= 224) && ($asciiPos <= 239)) // 3 chars representing one unicode character
  68          $i=3;
  69       else if (($asciiPos >= 192) && ($asciiPos <= 223)) // 2 chars representing one unicode character
  70          $i=2;
  71       else // 1 char (lower ascii)
  72          $i=1;
  73       $thisLetter = substr($source, $pos, $i);
  74       $pos += $i;
  75
  76       // process the string representing the letter to a unicode entity
  77       $thisLen = strlen($thisLetter);
  78       $thisPos = 0;
  79       $decimalCode = 0;
  80       while ($thisPos < $thisLen)
  81       {
  82          $thisCharOrd = ord(substr($thisLetter, $thisPos, 1));
  83          if ($thisPos == 0)
  84          {
  85             $charNum = intval($thisCharOrd - $decrement[$thisLen]);
  86             $decimalCode += ($charNum << $shift[$thisLen][$thisPos]);
  87          }
  88          else
  89          {
  90             $charNum = intval($thisCharOrd - 128);
  91             $decimalCode += ($charNum << $shift[$thisLen][$thisPos]);
  92          }
  93
  94          $thisPos++;
  95       }
  96
  97       $encodedLetter = '&#'. str_pad($decimalCode, ($thisLen==1)?3:5, '0', STR_PAD_LEFT).';';
  98       $encodedString .= $encodedLetter;
  99    }
 100
 101    return $encodedString;
 102 }
 103
 104 // vim:set et sw=4 sts=4 ts=4 enc=utf-8:
 105 ?>