[Summary view] [Print] [Text view]

   1  <?php
   2  /*
   3   * Limb PHP Framework
   4   *
   5   * @link http://limb-project.com 
   6   * @copyright  Copyright &copy; 2004-2007 BIT(http://bit-creative.com)
   7   * @license    LGPL http://www.gnu.org/copyleft/lesser.html 
   8   */
   9  lmb_require('limb/dbal/src/query/lmbInsertQuery.class.php');
  10  lmb_require('limb/dbal/src/query/lmbSelectQuery.class.php');
  11  lmb_require('limb/dbal/src/criteria/lmbSQLFieldCriteria.class.php');
  12  lmb_require('limb/dbal/src/query/lmbUpdateQuery.class.php');
  13  
  14  /**

  15   * class lmbFullTextSearchIndexer.

  16   *

  17   * @package search

  18   * @version $Id: lmbFullTextSearchIndexer.class.php 5945 2007-06-06 08:31:43Z pachanga $

  19   */
  20  class lmbFullTextSearchIndexer
  21  {
  22    protected $normalizer = null;
  23  
  24    protected $left_bound = '<!-- no index start -->';
  25    protected $right_bound = '<!-- no index end -->';
  26    protected $use_noindex = false;
  27  
  28    protected $conn;
  29  
  30    function __construct($normalizer)
  31    {
  32      $this->normalizer = $normalizer;
  33      $this->conn = lmbToolkit :: instance()->getDefaultDbConnection();
  34    }
  35  
  36    function useNOINDEX($status = true)
  37    {
  38      $this->use_noindex = $status;
  39    }
  40  
  41    function index($uri, $content)
  42    {
  43      $title = $this->_extractTitle($content);
  44      $content = $this->_getIndexedContent($content);
  45  
  46      $content = $this->normalizer->process($content);
  47  
  48      if($record = $this->findIndexRecordByUri($uri))
  49      {
  50        $this->_updateIndexRecordById($record->get('id'), $content, $title);
  51        return $record->get('id');
  52      }
  53      else
  54        return $this->_insertNewIndexRecord($uri, $content, $title);
  55    }
  56  
  57    function _getIndexedContent($content)
  58    {
  59      if(!$this->use_noindex)
  60        return $content;
  61  
  62      $regex = '~' .
  63               preg_quote($this->left_bound) .
  64               '(.*?)' .
  65               preg_quote($this->right_bound) .
  66               '~s';
  67  
  68      return preg_replace($regex, ' ', $content);
  69    }
  70  
  71    function _extractTitle(&$content)
  72    {
  73      $regex = '~<title>([^<]*)</title>~';
  74      if(preg_match($regex, $content, $matches))
  75        return $matches[1];
  76      else
  77        return '';
  78    }
  79  
  80    function _insertNewIndexRecord($uri, $content, $title)
  81    {
  82      $query = new lmbInsertQuery(FULL_TEXT_SEARCH_INDEXER_TABLE, $this->conn);
  83      $query->addField('uri', $uri->toString());
  84      $query->addField('content', $content);
  85      $query->addField('last_modified', time());
  86      $query->addField('title', $title);
  87      $stmt = $query->getStatement();
  88      return $stmt->insertId('id');
  89    }
  90  
  91    function _updateIndexRecordById($id, $content, $title)
  92    {
  93      $query = new lmbUpdateQuery(FULL_TEXT_SEARCH_INDEXER_TABLE, $this->conn);
  94      $query->addField('content', $content);
  95      $query->addField('last_modified', time());
  96      $query->addField('title', $title);
  97      $query->addCriteria(new lmbSQLFieldCriteria('id', $id));
  98      $stmt = $query->getStatement();
  99      $stmt->execute();
 100    }
 101  
 102    function findIndexRecordByUri($uri)
 103    {
 104      $query = new lmbSelectQuery(null, $this->conn);
 105      $query->addTable(FULL_TEXT_SEARCH_INDEXER_TABLE);
 106      $query->addCriteria(new lmbSQLFieldCriteria('uri', $uri->toString()));
 107      $rs = $query->getRecordSet();
 108      $rs->rewind();
 109      if($rs->valid())
 110        return $rs->current();
 111    }
 112  }
 113  
 114  ?>