Chris@76: <?php
Chris@76: 
Chris@76: /**
Chris@76:  * Simple Machines Forum (SMF)
Chris@76:  *
Chris@76:  * @package SMF
Chris@76:  * @author Simple Machines http://www.simplemachines.org
Chris@76:  * @copyright 2011 Simple Machines
Chris@76:  * @license http://www.simplemachines.org/about/smf/license.php BSD
Chris@76:  *
Chris@76:  * @version 2.0
Chris@76:  */
Chris@76: 
Chris@76: if (!defined('SMF'))
Chris@76: 	die('Hacking attempt...');
Chris@76: 
Chris@76: /*
Chris@76: 	int searchSort(string $wordA, string $wordB)
Chris@76: 		- callback function for usort used to sort the fulltext results.
Chris@76: 		- the order of sorting is: large words, small words, large words that
Chris@76: 		  are excluded from the search, small words that are excluded.
Chris@76: */
Chris@76: 
Chris@76: class custom_search
Chris@76: {
Chris@76: 	// This is the last version of SMF that this was tested on, to protect against API changes.
Chris@76: 	public $version_compatible = 'SMF 2.0';
Chris@76: 	// This won't work with versions of SMF less than this.
Chris@76: 	public $min_smf_version = 'SMF 2.0 Beta 2';
Chris@76: 	// Is it supported?
Chris@76: 	public $is_supported = true;
Chris@76: 
Chris@76: 	protected $indexSettings = array();
Chris@76: 	// What words are banned?
Chris@76: 	protected $bannedWords = array();
Chris@76: 	// What is the minimum word length?
Chris@76: 	protected $min_word_length = null;
Chris@76: 	// What databases support the custom index?
Chris@76: 	protected $supported_databases = array('mysql', 'postgresql', 'sqlite');
Chris@76: 
Chris@76: 	public function __construct()
Chris@76: 	{
Chris@76: 		global $modSettings, $db_type;
Chris@76: 
Chris@76: 		// Is this database supported?
Chris@76: 		if (!in_array($db_type, $this->supported_databases))
Chris@76: 		{
Chris@76: 			$this->is_supported = false;
Chris@76: 			return;
Chris@76: 		}
Chris@76: 
Chris@76: 		if (empty($modSettings['search_custom_index_config']))
Chris@76: 			return;
Chris@76: 
Chris@76: 		$this->indexSettings = unserialize($modSettings['search_custom_index_config']);
Chris@76: 
Chris@76: 		$this->bannedWords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
Chris@76: 		$this->min_word_length = $this->indexSettings['bytes_per_word'];
Chris@76: 	}
Chris@76: 
Chris@76: 	// Check whether the search can be performed by this API.
Chris@76: 	public function supportsMethod($methodName, $query_params = null)
Chris@76: 	{
Chris@76: 		switch ($methodName)
Chris@76: 		{
Chris@76: 			case 'isValid':
Chris@76: 			case 'searchSort':
Chris@76: 			case 'prepareIndexes':
Chris@76: 			case 'indexedWordQuery':
Chris@76: 				return true;
Chris@76: 			break;
Chris@76: 
Chris@76: 			default:
Chris@76: 
Chris@76: 				// All other methods, too bad dunno you.
Chris@76: 				return false;
Chris@76: 			return;
Chris@76: 		}
Chris@76: 	}
Chris@76: 
Chris@76: 	// If the settings don't exist we can't continue.
Chris@76: 	public function isValid()
Chris@76: 	{
Chris@76: 		global $modSettings;
Chris@76: 
Chris@76: 		return !empty($modSettings['search_custom_index_config']);
Chris@76: 	}
Chris@76: 
Chris@76: 	// This function compares the length of two strings plus a little.
Chris@76: 	public function searchSort($a, $b)
Chris@76: 	{
Chris@76: 		global $modSettings, $excludedWords;
Chris@76: 
Chris@76: 		$x = strlen($a) - (in_array($a, $excludedWords) ? 1000 : 0);
Chris@76: 		$y = strlen($b) - (in_array($b, $excludedWords) ? 1000 : 0);
Chris@76: 
Chris@76: 		return $y < $x ? 1 : ($y > $x ? -1 : 0);
Chris@76: 	}
Chris@76: 
Chris@76: 	// Do we have to do some work with the words we are searching for to prepare them?
Chris@76: 	public function prepareIndexes($word, &$wordsSearch, &$wordsExclude, $isExcluded)
Chris@76: 	{
Chris@76: 		global $modSettings, $smcFunc;
Chris@76: 
Chris@76: 		$subwords = text2words($word, $this->min_word_length, true);
Chris@76: 
Chris@76: 		if (empty($modSettings['search_force_index']))
Chris@76: 			$wordsSearch['words'][] = $word;
Chris@76: 
Chris@76: 		// Excluded phrases don't benefit from being split into subwords.
Chris@76: 		if (count($subwords) > 1 && $isExcluded)
Chris@76: 			continue;
Chris@76: 		else
Chris@76: 		{
Chris@76: 			foreach ($subwords as $subword)
Chris@76: 			{
Chris@76: 				if ($smcFunc['strlen']($subword) >= $this->min_word_length && !in_array($subword, $this->bannedWords))
Chris@76: 				{
Chris@76: 					$wordsSearch['indexed_words'][] = $subword;
Chris@76: 					if ($isExcluded)
Chris@76: 						$wordsExclude[] = $subword;
Chris@76: 				}
Chris@76: 			}
Chris@76: 		}
Chris@76: 	}
Chris@76: 
Chris@76: 	// Search for indexed words.
Chris@76: 	public function indexedWordQuery($words, $search_data)
Chris@76: 	{
Chris@76: 		global $modSettings, $smcFunc;
Chris@76: 
Chris@76: 		$query_select = array(
Chris@76: 			'id_msg' => 'm.id_msg',
Chris@76: 		);
Chris@76: 		$query_inner_join = array();
Chris@76: 		$query_left_join = array();
Chris@76: 		$query_where = array();
Chris@76: 		$query_params = $search_data['params'];
Chris@76: 
Chris@76: 		if ($query_params['id_search'])
Chris@76: 			$query_select['id_search'] = '{int:id_search}';
Chris@76: 
Chris@76: 		$count = 0;
Chris@76: 		foreach ($words['words'] as $regularWord)
Chris@76: 		{
Chris@76: 			$query_where[] = 'm.body' . (in_array($regularWord, $query_params['excluded_words']) ? ' NOT' : '') . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:complex_body_' . $count . '}';
Chris@76: 			$query_params['complex_body_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($regularWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $regularWord), '\\\'') . '[[:>:]]';
Chris@76: 		}
Chris@76: 
Chris@76: 		if ($query_params['user_query'])
Chris@76: 			$query_where[] = '{raw:user_query}';
Chris@76: 		if ($query_params['board_query'])
Chris@76: 			$query_where[] = 'm.id_board {raw:board_query}';
Chris@76: 
Chris@76: 		if ($query_params['topic'])
Chris@76: 			$query_where[] = 'm.id_topic = {int:topic}';
Chris@76: 		if ($query_params['min_msg_id'])
Chris@76: 			$query_where[] = 'm.id_msg >= {int:min_msg_id}';
Chris@76: 		if ($query_params['max_msg_id'])
Chris@76: 			$query_where[] = 'm.id_msg <= {int:max_msg_id}';
Chris@76: 
Chris@76: 		$count = 0;
Chris@76: 		if (!empty($query_params['excluded_phrases']) && empty($modSettings['search_force_index']))
Chris@76: 			foreach ($query_params['excluded_phrases'] as $phrase)
Chris@76: 			{
Chris@76: 				$query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_phrase_' . $count . '}';
Chris@76: 				$query_params['exclude_subject_phrase_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($phrase, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $phrase), '\\\'') . '[[:>:]]';
Chris@76: 			}
Chris@76: 		$count = 0;
Chris@76: 		if (!empty($query_params['excluded_subject_words']) && empty($modSettings['search_force_index']))
Chris@76: 			foreach ($query_params['excluded_subject_words'] as $excludedWord)
Chris@76: 			{
Chris@76: 				$query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_words_' . $count . '}';
Chris@76: 				$query_params['exclude_subject_words_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($excludedWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $excludedWord), '\\\'') . '[[:>:]]';
Chris@76: 			}
Chris@76: 
Chris@76: 		$numTables = 0;
Chris@76: 		$prev_join = 0;
Chris@76: 		foreach ($words['indexed_words'] as $indexedWord)
Chris@76: 		{
Chris@76: 			$numTables++;
Chris@76: 			if (in_array($indexedWord, $query_params['excluded_index_words']))
Chris@76: 			{
Chris@76: 				$query_left_join[] = '{db_prefix}log_search_words AS lsw' . $numTables . ' ON (lsw' . $numTables . '.id_word = ' . $indexedWord . ' AND lsw' . $numTables . '.id_msg = m.id_msg)';
Chris@76: 				$query_where[] = '(lsw' . $numTables . '.id_word IS NULL)';
Chris@76: 			}
Chris@76: 			else
Chris@76: 			{
Chris@76: 				$query_inner_join[] = '{db_prefix}log_search_words AS lsw' . $numTables . ' ON (lsw' . $numTables . '.id_msg = ' . ($prev_join === 0 ? 'm' : 'lsw' . $prev_join) . '.id_msg)';
Chris@76: 				$query_where[] = 'lsw' . $numTables . '.id_word = ' . $indexedWord;
Chris@76: 				$prev_join = $numTables;
Chris@76: 			}
Chris@76: 		}
Chris@76: 
Chris@76: 		$ignoreRequest = $smcFunc['db_search_query']('insert_into_log_messages_fulltext', ($smcFunc['db_support_ignore'] ? ( '
Chris@76: 			INSERT IGNORE INTO {db_prefix}' . $search_data['insert_into'] . '
Chris@76: 				(' . implode(', ', array_keys($query_select)) . ')') : '') . '
Chris@76: 			SELECT ' . implode(', ', $query_select) . '
Chris@76: 			FROM {db_prefix}messages AS m' . (empty($query_inner_join) ? '' : '
Chris@76: 				INNER JOIN ' . implode('
Chris@76: 				INNER JOIN ', $query_inner_join)) . (empty($query_left_join) ? '' : '
Chris@76: 				LEFT JOIN ' . implode('
Chris@76: 				LEFT JOIN ', $query_left_join)) . '
Chris@76: 			WHERE ' . implode('
Chris@76: 				AND ', $query_where) . (empty($search_data['max_results']) ? '' : '
Chris@76: 			LIMIT ' . ($search_data['max_results'] - $search_data['indexed_results'])),
Chris@76: 			$query_params
Chris@76: 		);
Chris@76: 
Chris@76: 		return $ignoreRequest;
Chris@76: 	}
Chris@76: }
Chris@76: 
Chris@76: ?>