utf8/substr.php

   1 <?php defined('SYSPATH') OR die('No direct script access.');
   2 /**
   3  * UTF8::substr
   4  *
   5  * @package    Kohana
   6  * @author     Kohana Team
   7  * @copyright  (c) 2007-2012 Kohana Team
   8  * @copyright  (c) 2005 Harry Fuecks
   9  * @license    http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt
  10  */
  11 function _substr($str, $offset, $length = NULL)
  12 {
  13         if (UTF8::is_ascii($str))
  14                 return ($length === NULL) ? substr($str, $offset) : substr($str, $offset, $length);
  15
  16         // Normalize params
  17         $str    = (string) $str;
  18         $strlen = UTF8::strlen($str);
  19         $offset = (int) ($offset < 0) ? max(0, $strlen + $offset) : $offset; // Normalize to positive offset
  20         $length = ($length === NULL) ? NULL : (int) $length;
  21
  22         // Impossible
  23         if ($length === 0 OR $offset >= $strlen OR ($length < 0 AND $length <= $offset - $strlen))
  24                 return '';
  25
  26         // Whole string
  27         if ($offset == 0 AND ($length === NULL OR $length >= $strlen))
  28                 return $str;
  29
  30         // Build regex
  31         $regex = '^';
  32
  33         // Create an offset expression
  34         if ($offset > 0)
  35         {
  36                 // PCRE repeating quantifiers must be less than 65536, so repeat when necessary
  37                 $x = (int) ($offset / 65535);
  38                 $y = (int) ($offset % 65535);
  39                 $regex .= ($x == 0) ? '' : ('(?:.{65535}){'.$x.'}');
  40                 $regex .= ($y == 0) ? '' : ('.{'.$y.'}');
  41         }
  42
  43         // Create a length expression
  44         if ($length === NULL)
  45         {
  46                 $regex .= '(.*)'; // No length set, grab it all
  47         }
  48         // Find length from the left (positive length)
  49         elseif ($length > 0)
  50         {
  51                 // Reduce length so that it can't go beyond the end of the string
  52                 $length = min($strlen - $offset, $length);
  53
  54                 $x = (int) ($length / 65535);
  55                 $y = (int) ($length % 65535);
  56                 $regex .= '(';
  57                 $regex .= ($x == 0) ? '' : ('(?:.{65535}){'.$x.'}');
  58                 $regex .= '.{'.$y.'})';
  59         }
  60         // Find length from the right (negative length)
  61         else
  62         {
  63                 $x = (int) (-$length / 65535);
  64                 $y = (int) (-$length % 65535);
  65                 $regex .= '(.*)';
  66                 $regex .= ($x == 0) ? '' : ('(?:.{65535}){'.$x.'}');
  67                 $regex .= '.{'.$y.'}';
  68         }
  69
  70         preg_match('/'.$regex.'/us', $str, $matches);
  71         return $matches[1];
  72 }