opNGramSearchTextAnalyzer.class.php

<?php

/**
 * This file is part of the OpenPNE package.
 * (c) OpenPNE Project (http://www.openpne.jp/)
 *
 * For the full copyright and license information, please view the LICENSE
 * file and the NOTICE file that were distributed with this source code.
 */

/**
 * opNGramSearchTextAnalyzer
 *
 * @package    OpenPNE
 * @subpackage util
 * @author     Kousuke Ebihara <ebihara@tejimaya.com>
 */
class opNGramSearchTextAnalyzer extends Doctrine_Search_Analyzer_Standard
{
  public function analyze($text)
  {
    $result = array();

    $text = preg_replace('/[^\x21-\x7e]+/u', '', $text);


    if (!$text)
    {
      return $result;
    }

    $length = Doctrine_Validator::getStringLength($text);
    for ($i = 0; $i < $length; $i++)
    {
      $result[] = $this->substrNGram($text, $i, 2);
    }

    return array_unique($result);
  }

  protected function substrNGram($string, $offset, $length)
  {
    if (function_exists('iconv_substr'))
    {
      return iconv_substr($string, $offset, $length, 'UTF-8');
    }
    elseif (function_exists('mb_substr'))
    {
      return mb_substr($string, $offset, $length, 'UTF-8');
    }

    return substr(utf8_decode($string), $offset, $length);
  }
}