123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281 |
- <?php
- /**
- * Simple Machines Forum (SMF)
- *
- * @package SMF
- * @author Simple Machines http://www.simplemachines.org
- * @copyright 2011 Simple Machines
- * @license http://www.simplemachines.org/about/smf/license.php BSD
- *
- * @version 2.1 Alpha 1
- */
- if (!defined('SMF'))
- die('Hacking attempt...');
- class custom_search
- {
- // This is the last version of SMF that this was tested on, to protect against API changes.
- public $version_compatible = 'SMF 2.1 Alpha 1';
- // This won't work with versions of SMF less than this.
- public $min_smf_version = 'SMF 2.1 Alpha 1';
- // Is it supported?
- public $is_supported = true;
- protected $indexSettings = array();
- // What words are banned?
- protected $bannedWords = array();
- // What is the minimum word length?
- protected $min_word_length = null;
- // What databases support the custom index?
- protected $supported_databases = array('mysql', 'postgresql', 'sqlite');
- public function __construct()
- {
- global $modSettings, $db_type;
- // Is this database supported?
- if (!in_array($db_type, $this->supported_databases))
- {
- $this->is_supported = false;
- return;
- }
- if (empty($modSettings['search_custom_index_config']))
- return;
- $this->indexSettings = unserialize($modSettings['search_custom_index_config']);
- $this->bannedWords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
- $this->min_word_length = $this->indexSettings['bytes_per_word'];
- }
- // Check whether the search can be performed by this API.
- public function supportsMethod($methodName, $query_params = null)
- {
- switch ($methodName)
- {
- case 'isValid':
- case 'searchSort':
- case 'prepareIndexes':
- case 'indexedWordQuery':
- case 'postCreated':
- case 'postModified':
- return true;
- break;
- // All other methods, too bad dunno you.
- default:
- return false;
- return;
- }
- }
- // If the settings don't exist we can't continue.
- public function isValid()
- {
- global $modSettings;
- return !empty($modSettings['search_custom_index_config']);
- }
- /**
- * callback function for usort used to sort the fulltext results.
- * the order of sorting is: large words, small words, large words that
- * are excluded from the search, small words that are excluded.
- * @param string $a Word A
- * @param string $b Word B
- * @return int
- */
- public function searchSort($a, $b)
- {
- global $modSettings, $excludedWords;
- $x = strlen($a) - (in_array($a, $excludedWords) ? 1000 : 0);
- $y = strlen($b) - (in_array($b, $excludedWords) ? 1000 : 0);
- return $y < $x ? 1 : ($y > $x ? -1 : 0);
- }
- // Do we have to do some work with the words we are searching for to prepare them?
- public function prepareIndexes($word, &$wordsSearch, &$wordsExclude, $isExcluded)
- {
- global $modSettings, $smcFunc;
- $subwords = text2words($word, $this->min_word_length, true);
- if (empty($modSettings['search_force_index']))
- $wordsSearch['words'][] = $word;
- // Excluded phrases don't benefit from being split into subwords.
- if (count($subwords) > 1 && $isExcluded)
- continue;
- else
- {
- foreach ($subwords as $subword)
- {
- if ($smcFunc['strlen']($subword) >= $this->min_word_length && !in_array($subword, $this->bannedWords))
- {
- $wordsSearch['indexed_words'][] = $subword;
- if ($isExcluded)
- $wordsExclude[] = $subword;
- }
- }
- }
- }
- // Search for indexed words.
- public function indexedWordQuery($words, $search_data)
- {
- global $modSettings, $smcFunc;
- $query_select = array(
- 'id_msg' => 'm.id_msg',
- );
- $query_inner_join = array();
- $query_left_join = array();
- $query_where = array();
- $query_params = $search_data['params'];
- if ($query_params['id_search'])
- $query_select['id_search'] = '{int:id_search}';
- $count = 0;
- foreach ($words['words'] as $regularWord)
- {
- $query_where[] = 'm.body' . (in_array($regularWord, $query_params['excluded_words']) ? ' NOT' : '') . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:complex_body_' . $count . '}';
- $query_params['complex_body_' . $count++] = empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? '%' . strtr($regularWord, array('_' => '\\_', '%' => '\\%')) . '%' : '[[:<:]]' . addcslashes(preg_replace(array('/([\[\]$.+*?|{}()])/'), array('[$1]'), $regularWord), '\\\'') . '[[:>:]]';
- }
- if ($query_params['user_query'])
- $query_where[] = '{raw:user_query}';
- if ($query_params['board_query'])
- $query_where[] = 'm.id_board {raw:board_query}';
- if ($query_params['topic'])
- $query_where[] = 'm.id_topic = {int:topic}';
- if ($query_params['min_msg_id'])
- $query_where[] = 'm.id_msg >= {int:min_msg_id}';
- if ($query_params['max_msg_id'])
- $query_where[] = 'm.id_msg <= {int:max_msg_id}';
- $count = 0;
- if (!empty($query_params['excluded_phrases']) && empty($modSettings['search_force_index']))
- foreach ($query_params['excluded_phrases'] as $phrase)
- {
- $query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_phrase_' . $count . '}';
- $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), '\\\'') . '[[:>:]]';
- }
- $count = 0;
- if (!empty($query_params['excluded_subject_words']) && empty($modSettings['search_force_index']))
- foreach ($query_params['excluded_subject_words'] as $excludedWord)
- {
- $query_where[] = 'subject NOT ' . (empty($modSettings['search_match_words']) || $search_data['no_regexp'] ? ' LIKE ' : ' RLIKE ') . '{string:exclude_subject_words_' . $count . '}';
- $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), '\\\'') . '[[:>:]]';
- }
- $numTables = 0;
- $prev_join = 0;
- foreach ($words['indexed_words'] as $indexedWord)
- {
- $numTables++;
- if (in_array($indexedWord, $query_params['excluded_index_words']))
- {
- $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)';
- $query_where[] = '(lsw' . $numTables . '.id_word IS NULL)';
- }
- else
- {
- $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)';
- $query_where[] = 'lsw' . $numTables . '.id_word = ' . $indexedWord;
- $prev_join = $numTables;
- }
- }
- $ignoreRequest = $smcFunc['db_search_query']('insert_into_log_messages_fulltext', ($smcFunc['db_support_ignore'] ? ( '
- INSERT IGNORE INTO {db_prefix}' . $search_data['insert_into'] . '
- (' . implode(', ', array_keys($query_select)) . ')') : '') . '
- SELECT ' . implode(', ', $query_select) . '
- FROM {db_prefix}messages AS m' . (empty($query_inner_join) ? '' : '
- INNER JOIN ' . implode('
- INNER JOIN ', $query_inner_join)) . (empty($query_left_join) ? '' : '
- LEFT JOIN ' . implode('
- LEFT JOIN ', $query_left_join)) . '
- WHERE ' . implode('
- AND ', $query_where) . (empty($search_data['max_results']) ? '' : '
- LIMIT ' . ($search_data['max_results'] - $search_data['indexed_results'])),
- $query_params
- );
- return $ignoreRequest;
- }
- /**
- * After a post is made, we update the database.
- */
- public function postCreated($msgOptions, $topicOptions, $posterOptions)
- {
- global $modSettings, $smcFunc;
- $customIndexSettings = unserialize($modSettings['search_custom_index_config']);
- $inserts = array();
- foreach (text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true) as $word)
- $inserts[] = array($word, $msgOptions['id']);
- if (!empty($inserts))
- $smcFunc['db_insert']('ignore',
- '{db_prefix}log_search_words',
- array('id_word' => 'int', 'id_msg' => 'int'),
- $inserts,
- array('id_word', 'id_msg')
- );
- }
- /**
- * After a post is modified, we update the database.
- */
- public function postModified($msgOptions, $topicOptions, $posterOptions)
- {
- global $modSettings, $smcFunc;
- $customIndexSettings = unserialize($modSettings['search_custom_index_config']);
- $stopwords = empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
- $old_index = text2words($msgOptions['old_body'], $customIndexSettings['bytes_per_word'], true);
- $new_index = text2words($msgOptions['body'], $customIndexSettings['bytes_per_word'], true);
- // Calculate the words to be added and removed from the index.
- $removed_words = array_diff(array_diff($old_index, $new_index), $stopwords);
- $inserted_words = array_diff(array_diff($new_index, $old_index), $stopwords);
- // Delete the removed words AND the added ones to avoid key constraints.
- if (!empty($removed_words))
- {
- $removed_words = array_merge($removed_words, $inserted_words);
- $smcFunc['db_query']('', '
- DELETE FROM {db_prefix}log_search_words
- WHERE id_msg = {int:id_msg}
- AND id_word IN ({array_int:removed_words})',
- array(
- 'removed_words' => $removed_words,
- 'id_msg' => $msgOptions['id'],
- )
- );
- }
- // Add the new words to be indexed.
- if (!empty($inserted_words))
- {
- $inserts = array();
- foreach ($inserted_words as $word)
- $inserts[] = array($word, $msgOptions['id']);
- $smcFunc['db_insert']('insert',
- '{db_prefix}log_search_words',
- array('id_word' => 'string', 'id_msg' => 'int'),
- $inserts,
- array('id_word', 'id_msg')
- );
- }
- }
- }
|