ManageSearch.php 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800
  1. <?php
  2. /**
  3. * The admin screen to change the search settings.
  4. *
  5. * Simple Machines Forum (SMF)
  6. *
  7. * @package SMF
  8. * @author Simple Machines http://www.simplemachines.org
  9. * @copyright 2013 Simple Machines and individual contributors
  10. * @license http://www.simplemachines.org/about/smf/license.php BSD
  11. *
  12. * @version 2.1 Alpha 1
  13. */
  14. if (!defined('SMF'))
  15. die('No direct access...');
  16. /**
  17. * Main entry point for the admin search settings screen.
  18. * It checks permissions, and it forwards to the appropriate function based on
  19. * the given sub-action.
  20. * Defaults to sub-action 'settings'.
  21. * Called by ?action=admin;area=managesearch.
  22. * Requires the admin_forum permission.
  23. *
  24. * @uses ManageSearch template.
  25. * @uses Search language file.
  26. */
  27. function ManageSearch()
  28. {
  29. global $context, $txt, $scripturl;
  30. isAllowedTo('admin_forum');
  31. loadLanguage('Search');
  32. loadTemplate('ManageSearch');
  33. db_extend('search');
  34. $subActions = array(
  35. 'settings' => 'EditSearchSettings',
  36. 'weights' => 'EditWeights',
  37. 'method' => 'EditSearchMethod',
  38. 'createfulltext' => 'EditSearchMethod',
  39. 'removecustom' => 'EditSearchMethod',
  40. 'removefulltext' => 'EditSearchMethod',
  41. 'createmsgindex' => 'CreateMessageIndex',
  42. );
  43. call_integration_hook('integrate_manage_search', array(&$subActions));
  44. // Default the sub-action to 'edit search settings'.
  45. $_REQUEST['sa'] = isset($_REQUEST['sa']) && isset($subActions[$_REQUEST['sa']]) ? $_REQUEST['sa'] : 'weights';
  46. $context['sub_action'] = $_REQUEST['sa'];
  47. // Create the tabs for the template.
  48. $context[$context['admin_menu_name']]['tab_data'] = array(
  49. 'title' => $txt['manage_search'],
  50. 'help' => 'search',
  51. 'description' => $txt['search_settings_desc'],
  52. 'tabs' => array(
  53. 'weights' => array(
  54. 'description' => $txt['search_weights_desc'],
  55. ),
  56. 'method' => array(
  57. 'description' => $txt['search_method_desc'],
  58. ),
  59. 'settings' => array(
  60. 'description' => $txt['search_settings_desc'],
  61. ),
  62. ),
  63. );
  64. // Call the right function for this sub-action.
  65. $subActions[$_REQUEST['sa']]();
  66. }
  67. /**
  68. * Edit some general settings related to the search function.
  69. * Called by ?action=admin;area=managesearch;sa=settings.
  70. * Requires the admin_forum permission.
  71. *
  72. * @param $return_config
  73. * @uses ManageSearch template, 'modify_settings' sub-template.
  74. */
  75. function EditSearchSettings($return_config = false)
  76. {
  77. global $txt, $context, $scripturl, $sourcedir, $modSettings;
  78. // What are we editing anyway?
  79. $config_vars = array(
  80. // Permission...
  81. array('permissions', 'search_posts'),
  82. // Some simple settings.
  83. array('check', 'simpleSearch'),
  84. array('check', 'search_dropdown'),
  85. array('int', 'search_results_per_page'),
  86. array('int', 'search_max_results', 'subtext' => $txt['search_max_results_disable']),
  87. '',
  88. // Some limitations.
  89. array('int', 'search_floodcontrol_time', 'subtext' => $txt['search_floodcontrol_time_desc'], 6, 'postinput' => $txt['seconds']),
  90. );
  91. call_integration_hook('integrate_modify_search_settings', array(&$config_vars));
  92. // Perhaps the search method wants to add some settings?
  93. require_once($sourcedir . '/Search.php');
  94. $searchAPI = findSearchAPI();
  95. if (is_callable(array($searchAPI, 'searchSettings')))
  96. call_user_func_array($searchAPI->searchSettings, array(&$config_vars));
  97. if ($return_config)
  98. return $config_vars;
  99. $context['page_title'] = $txt['search_settings_title'];
  100. $context['sub_template'] = 'show_settings';
  101. // We'll need this for the settings.
  102. require_once($sourcedir . '/ManageServer.php');
  103. // A form was submitted.
  104. if (isset($_REQUEST['save']))
  105. {
  106. checkSession();
  107. call_integration_hook('integrate_save_search_settings');
  108. saveDBSettings($config_vars);
  109. redirectexit('action=admin;area=managesearch;sa=settings;' . $context['session_var'] . '=' . $context['session_id']);
  110. }
  111. // Prep the template!
  112. $context['post_url'] = $scripturl . '?action=admin;area=managesearch;save;sa=settings';
  113. $context['settings_title'] = $txt['search_settings_title'];
  114. // We need this for the in-line permissions
  115. createToken('admin-mp');
  116. prepareDBSettingContext($config_vars);
  117. }
  118. /**
  119. * Edit the relative weight of the search factors.
  120. * Called by ?action=admin;area=managesearch;sa=weights.
  121. * Requires the admin_forum permission.
  122. *
  123. * @uses ManageSearch template, 'modify_weights' sub-template.
  124. */
  125. function EditWeights()
  126. {
  127. global $txt, $context, $modSettings;
  128. $context['page_title'] = $txt['search_weights_title'];
  129. $context['sub_template'] = 'modify_weights';
  130. $factors = array(
  131. 'search_weight_frequency',
  132. 'search_weight_age',
  133. 'search_weight_length',
  134. 'search_weight_subject',
  135. 'search_weight_first_message',
  136. 'search_weight_sticky',
  137. );
  138. call_integration_hook('integrate_modify_search_weights', array(&$factors));
  139. // A form was submitted.
  140. if (isset($_POST['save']))
  141. {
  142. checkSession();
  143. validateToken('admin-msw');
  144. call_integration_hook('integrate_save_search_weights');
  145. $changes = array();
  146. foreach ($factors as $factor)
  147. $changes[$factor] = (int) $_POST[$factor];
  148. updateSettings($changes);
  149. }
  150. $context['relative_weights'] = array('total' => 0);
  151. foreach ($factors as $factor)
  152. $context['relative_weights']['total'] += isset($modSettings[$factor]) ? $modSettings[$factor] : 0;
  153. foreach ($factors as $factor)
  154. $context['relative_weights'][$factor] = round(100 * (isset($modSettings[$factor]) ? $modSettings[$factor] : 0) / $context['relative_weights']['total'], 1);
  155. createToken('admin-msw');
  156. }
  157. /**
  158. * Edit the search method and search index used.
  159. * Calculates the size of the current search indexes in use.
  160. * Allows to create and delete a fulltext index on the messages table.
  161. * Allows to delete a custom index (that CreateMessageIndex() created).
  162. * Called by ?action=admin;area=managesearch;sa=method.
  163. * Requires the admin_forum permission.
  164. *
  165. * @uses ManageSearch template, 'select_search_method' sub-template.
  166. */
  167. function EditSearchMethod()
  168. {
  169. global $txt, $context, $modSettings, $smcFunc, $db_type, $db_prefix;
  170. $context[$context['admin_menu_name']]['current_subsection'] = 'method';
  171. $context['page_title'] = $txt['search_method_title'];
  172. $context['sub_template'] = 'select_search_method';
  173. $context['supports_fulltext'] = $smcFunc['db_search_support']('fulltext');
  174. // Load any apis.
  175. $context['search_apis'] = loadSearchAPIs();
  176. // Detect whether a fulltext index is set.
  177. if ($context['supports_fulltext'])
  178. detectFulltextIndex();
  179. if (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'createfulltext')
  180. {
  181. checkSession('get');
  182. validateToken('admin-msm', 'get');
  183. // Make sure it's gone before creating it.
  184. $smcFunc['db_query']('', '
  185. ALTER TABLE {db_prefix}messages
  186. DROP INDEX body',
  187. array(
  188. 'db_error_skip' => true,
  189. )
  190. );
  191. $smcFunc['db_query']('', '
  192. ALTER TABLE {db_prefix}messages
  193. ADD FULLTEXT body (body)',
  194. array(
  195. )
  196. );
  197. $context['fulltext_index'] = 'body';
  198. }
  199. elseif (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'removefulltext' && !empty($context['fulltext_index']))
  200. {
  201. checkSession('get');
  202. validateToken('admin-msm', 'get');
  203. $smcFunc['db_query']('', '
  204. ALTER TABLE {db_prefix}messages
  205. DROP INDEX ' . implode(',
  206. DROP INDEX ', $context['fulltext_index']),
  207. array(
  208. 'db_error_skip' => true,
  209. )
  210. );
  211. $context['fulltext_index'] = '';
  212. // Go back to the default search method.
  213. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'fulltext')
  214. updateSettings(array(
  215. 'search_index' => '',
  216. ));
  217. }
  218. elseif (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'removecustom')
  219. {
  220. checkSession('get');
  221. validateToken('admin-msm', 'get');
  222. db_extend();
  223. $tables = $smcFunc['db_list_tables'](false, $db_prefix . 'log_search_words');
  224. if (!empty($tables))
  225. {
  226. $smcFunc['db_search_query']('drop_words_table', '
  227. DROP TABLE {db_prefix}log_search_words',
  228. array(
  229. )
  230. );
  231. }
  232. updateSettings(array(
  233. 'search_custom_index_config' => '',
  234. 'search_custom_index_resume' => '',
  235. ));
  236. // Go back to the default search method.
  237. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'custom')
  238. updateSettings(array(
  239. 'search_index' => '',
  240. ));
  241. }
  242. elseif (isset($_POST['save']))
  243. {
  244. checkSession();
  245. validateToken('admin-msmpost');
  246. updateSettings(array(
  247. 'search_index' => empty($_POST['search_index']) || (!in_array($_POST['search_index'], array('fulltext', 'custom')) && !isset($context['search_apis'][$_POST['search_index']])) ? '' : $_POST['search_index'],
  248. 'search_force_index' => isset($_POST['search_force_index']) ? '1' : '0',
  249. 'search_match_words' => isset($_POST['search_match_words']) ? '1' : '0',
  250. ));
  251. }
  252. $context['table_info'] = array(
  253. 'data_length' => 0,
  254. 'index_length' => 0,
  255. 'fulltext_length' => 0,
  256. 'custom_index_length' => 0,
  257. );
  258. // Get some info about the messages table, to show its size and index size.
  259. if ($db_type == 'mysql' || $db_type == 'mysqli')
  260. {
  261. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  262. $request = $smcFunc['db_query']('', '
  263. SHOW TABLE STATUS
  264. FROM {string:database_name}
  265. LIKE {string:table_name}',
  266. array(
  267. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  268. 'table_name' => str_replace('_', '\_', $match[2]) . 'messages',
  269. )
  270. );
  271. else
  272. $request = $smcFunc['db_query']('', '
  273. SHOW TABLE STATUS
  274. LIKE {string:table_name}',
  275. array(
  276. 'table_name' => str_replace('_', '\_', $db_prefix) . 'messages',
  277. )
  278. );
  279. if ($request !== false && $smcFunc['db_num_rows']($request) == 1)
  280. {
  281. // Only do this if the user has permission to execute this query.
  282. $row = $smcFunc['db_fetch_assoc']($request);
  283. $context['table_info']['data_length'] = $row['Data_length'];
  284. $context['table_info']['index_length'] = $row['Index_length'];
  285. $context['table_info']['fulltext_length'] = $row['Index_length'];
  286. $smcFunc['db_free_result']($request);
  287. }
  288. // Now check the custom index table, if it exists at all.
  289. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  290. $request = $smcFunc['db_query']('', '
  291. SHOW TABLE STATUS
  292. FROM {string:database_name}
  293. LIKE {string:table_name}',
  294. array(
  295. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  296. 'table_name' => str_replace('_', '\_', $match[2]) . 'log_search_words',
  297. )
  298. );
  299. else
  300. $request = $smcFunc['db_query']('', '
  301. SHOW TABLE STATUS
  302. LIKE {string:table_name}',
  303. array(
  304. 'table_name' => str_replace('_', '\_', $db_prefix) . 'log_search_words',
  305. )
  306. );
  307. if ($request !== false && $smcFunc['db_num_rows']($request) == 1)
  308. {
  309. // Only do this if the user has permission to execute this query.
  310. $row = $smcFunc['db_fetch_assoc']($request);
  311. $context['table_info']['index_length'] += $row['Data_length'] + $row['Index_length'];
  312. $context['table_info']['custom_index_length'] = $row['Data_length'] + $row['Index_length'];
  313. $smcFunc['db_free_result']($request);
  314. }
  315. }
  316. elseif ($db_type == 'postgresql')
  317. {
  318. // In order to report the sizes correctly we need to perform vacuum (optimize) on the tables we will be using.
  319. db_extend();
  320. $temp_tables = $smcFunc['db_list_tables']();
  321. foreach ($temp_tables as $table)
  322. if ($table == $db_prefix. 'messages' || $table == $db_prefix. 'log_search_words')
  323. $smcFunc['db_optimize_table']($table);
  324. // PostGreSql has some hidden sizes.
  325. $request = $smcFunc['db_query']('', '
  326. SELECT relname, relpages * 8 *1024 AS "KB" FROM pg_class
  327. WHERE relname = {string:messages} OR relname = {string:log_search_words}
  328. ORDER BY relpages DESC',
  329. array(
  330. 'messages' => $db_prefix. 'messages',
  331. 'log_search_words' => $db_prefix. 'log_search_words',
  332. )
  333. );
  334. if ($request !== false && $smcFunc['db_num_rows']($request) > 0)
  335. {
  336. while ($row = $smcFunc['db_fetch_assoc']($request))
  337. {
  338. if ($row['relname'] == $db_prefix . 'messages')
  339. {
  340. $context['table_info']['data_length'] = (int) $row['KB'];
  341. $context['table_info']['index_length'] = (int) $row['KB'];
  342. // Doesn't support fulltext
  343. $context['table_info']['fulltext_length'] = $txt['not_applicable'];
  344. }
  345. elseif ($row['relname'] == $db_prefix. 'log_search_words')
  346. {
  347. $context['table_info']['index_length'] = (int) $row['KB'];
  348. $context['table_info']['custom_index_length'] = (int) $row['KB'];
  349. }
  350. }
  351. $smcFunc['db_free_result']($request);
  352. }
  353. else
  354. // Didn't work for some reason...
  355. $context['table_info'] = array(
  356. 'data_length' => $txt['not_applicable'],
  357. 'index_length' => $txt['not_applicable'],
  358. 'fulltext_length' => $txt['not_applicable'],
  359. 'custom_index_length' => $txt['not_applicable'],
  360. );
  361. }
  362. else
  363. $context['table_info'] = array(
  364. 'data_length' => $txt['not_applicable'],
  365. 'index_length' => $txt['not_applicable'],
  366. 'fulltext_length' => $txt['not_applicable'],
  367. 'custom_index_length' => $txt['not_applicable'],
  368. );
  369. // Format the data and index length in kilobytes.
  370. foreach ($context['table_info'] as $type => $size)
  371. {
  372. // If it's not numeric then just break. This database engine doesn't support size.
  373. if (!is_numeric($size))
  374. break;
  375. $context['table_info'][$type] = comma_format($context['table_info'][$type] / 1024) . ' ' . $txt['search_method_kilobytes'];
  376. }
  377. $context['custom_index'] = !empty($modSettings['search_custom_index_config']);
  378. $context['partial_custom_index'] = !empty($modSettings['search_custom_index_resume']) && empty($modSettings['search_custom_index_config']);
  379. $context['double_index'] = !empty($context['fulltext_index']) && $context['custom_index'];
  380. createToken('admin-msmpost');
  381. createToken('admin-msm', 'get');
  382. }
  383. /**
  384. * Create a custom search index for the messages table.
  385. * Called by ?action=admin;area=managesearch;sa=createmsgindex.
  386. * Linked from the EditSearchMethod screen.
  387. * Requires the admin_forum permission.
  388. * Depending on the size of the message table, the process is divided in steps.
  389. *
  390. * @uses ManageSearch template, 'create_index', 'create_index_progress', and 'create_index_done'
  391. * sub-templates.
  392. */
  393. function CreateMessageIndex()
  394. {
  395. global $modSettings, $context, $smcFunc, $db_prefix, $txt;
  396. // Scotty, we need more time...
  397. @set_time_limit(600);
  398. if (function_exists('apache_reset_timeout'))
  399. @apache_reset_timeout();
  400. $context[$context['admin_menu_name']]['current_subsection'] = 'method';
  401. $context['page_title'] = $txt['search_index_custom'];
  402. $messages_per_batch = 50;
  403. $index_properties = array(
  404. 2 => array(
  405. 'column_definition' => 'small',
  406. 'step_size' => 1000000,
  407. ),
  408. 4 => array(
  409. 'column_definition' => 'medium',
  410. 'step_size' => 1000000,
  411. 'max_size' => 16777215,
  412. ),
  413. 5 => array(
  414. 'column_definition' => 'large',
  415. 'step_size' => 100000000,
  416. 'max_size' => 2000000000,
  417. ),
  418. );
  419. if (isset($_REQUEST['resume']) && !empty($modSettings['search_custom_index_resume']))
  420. {
  421. $context['index_settings'] = unserialize($modSettings['search_custom_index_resume']);
  422. $context['start'] = (int) $context['index_settings']['resume_at'];
  423. unset($context['index_settings']['resume_at']);
  424. $context['step'] = 1;
  425. }
  426. else
  427. {
  428. $context['index_settings'] = array(
  429. 'bytes_per_word' => isset($_REQUEST['bytes_per_word']) && isset($index_properties[$_REQUEST['bytes_per_word']]) ? (int) $_REQUEST['bytes_per_word'] : 2,
  430. );
  431. $context['start'] = isset($_REQUEST['start']) ? (int) $_REQUEST['start'] : 0;
  432. $context['step'] = isset($_REQUEST['step']) ? (int) $_REQUEST['step'] : 0;
  433. // admin timeouts are painful when building these long indexes
  434. if ($_SESSION['admin_time'] + 3300 < time() && $context['step'] >= 1)
  435. $_SESSION['admin_time'] = time();
  436. }
  437. if ($context['step'] !== 0)
  438. checkSession('request');
  439. // Step 0: let the user determine how they like their index.
  440. if ($context['step'] === 0)
  441. {
  442. $context['sub_template'] = 'create_index';
  443. }
  444. // Step 1: insert all the words.
  445. if ($context['step'] === 1)
  446. {
  447. $context['sub_template'] = 'create_index_progress';
  448. if ($context['start'] === 0)
  449. {
  450. db_extend();
  451. $tables = $smcFunc['db_list_tables'](false, $db_prefix . 'log_search_words');
  452. if (!empty($tables))
  453. {
  454. $smcFunc['db_search_query']('drop_words_table', '
  455. DROP TABLE {db_prefix}log_search_words',
  456. array(
  457. )
  458. );
  459. }
  460. $smcFunc['db_create_word_search']($index_properties[$context['index_settings']['bytes_per_word']]['column_definition']);
  461. // Temporarily switch back to not using a search index.
  462. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'custom')
  463. updateSettings(array('search_index' => ''));
  464. // Don't let simultanious processes be updating the search index.
  465. if (!empty($modSettings['search_custom_index_config']))
  466. updateSettings(array('search_custom_index_config' => ''));
  467. }
  468. $num_messages = array(
  469. 'done' => 0,
  470. 'todo' => 0,
  471. );
  472. $request = $smcFunc['db_query']('', '
  473. SELECT id_msg >= {int:starting_id} AS todo, COUNT(*) AS num_messages
  474. FROM {db_prefix}messages
  475. GROUP BY todo',
  476. array(
  477. 'starting_id' => $context['start'],
  478. )
  479. );
  480. while ($row = $smcFunc['db_fetch_assoc']($request))
  481. $num_messages[empty($row['todo']) ? 'done' : 'todo'] = $row['num_messages'];
  482. if (empty($num_messages['todo']))
  483. {
  484. $context['step'] = 2;
  485. $context['percentage'] = 80;
  486. $context['start'] = 0;
  487. }
  488. else
  489. {
  490. // Number of seconds before the next step.
  491. $stop = time() + 3;
  492. while (time() < $stop)
  493. {
  494. $inserts = array();
  495. $request = $smcFunc['db_query']('', '
  496. SELECT id_msg, body
  497. FROM {db_prefix}messages
  498. WHERE id_msg BETWEEN {int:starting_id} AND {int:ending_id}
  499. LIMIT {int:limit}',
  500. array(
  501. 'starting_id' => $context['start'],
  502. 'ending_id' => $context['start'] + $messages_per_batch - 1,
  503. 'limit' => $messages_per_batch,
  504. )
  505. );
  506. $forced_break = false;
  507. $number_processed = 0;
  508. while ($row = $smcFunc['db_fetch_assoc']($request))
  509. {
  510. // In theory it's possible for one of these to take friggin ages so add more timeout protection.
  511. if ($stop < time())
  512. {
  513. $forced_break = true;
  514. break;
  515. }
  516. $number_processed++;
  517. foreach (text2words($row['body'], $context['index_settings']['bytes_per_word'], true) as $id_word)
  518. {
  519. $inserts[] = array($id_word, $row['id_msg']);
  520. }
  521. }
  522. $num_messages['done'] += $number_processed;
  523. $num_messages['todo'] -= $number_processed;
  524. $smcFunc['db_free_result']($request);
  525. $context['start'] += $forced_break ? $number_processed : $messages_per_batch;
  526. if (!empty($inserts))
  527. $smcFunc['db_insert']('ignore',
  528. '{db_prefix}log_search_words',
  529. array('id_word' => 'int', 'id_msg' => 'int'),
  530. $inserts,
  531. array('id_word', 'id_msg')
  532. );
  533. if ($num_messages['todo'] === 0)
  534. {
  535. $context['step'] = 2;
  536. $context['start'] = 0;
  537. break;
  538. }
  539. else
  540. updateSettings(array('search_custom_index_resume' => serialize(array_merge($context['index_settings'], array('resume_at' => $context['start'])))));
  541. }
  542. // Since there are still two steps to go, 80% is the maximum here.
  543. $context['percentage'] = round($num_messages['done'] / ($num_messages['done'] + $num_messages['todo']), 3) * 80;
  544. }
  545. }
  546. // Step 2: removing the words that occur too often and are of no use.
  547. elseif ($context['step'] === 2)
  548. {
  549. if ($context['index_settings']['bytes_per_word'] < 4)
  550. $context['step'] = 3;
  551. else
  552. {
  553. $stop_words = $context['start'] === 0 || empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
  554. $stop = time() + 3;
  555. $context['sub_template'] = 'create_index_progress';
  556. $max_messages = ceil(60 * $modSettings['totalMessages'] / 100);
  557. while (time() < $stop)
  558. {
  559. $request = $smcFunc['db_query']('', '
  560. SELECT id_word, COUNT(id_word) AS num_words
  561. FROM {db_prefix}log_search_words
  562. WHERE id_word BETWEEN {int:starting_id} AND {int:ending_id}
  563. GROUP BY id_word
  564. HAVING COUNT(id_word) > {int:minimum_messages}',
  565. array(
  566. 'starting_id' => $context['start'],
  567. 'ending_id' => $context['start'] + $index_properties[$context['index_settings']['bytes_per_word']]['step_size'] - 1,
  568. 'minimum_messages' => $max_messages,
  569. )
  570. );
  571. while ($row = $smcFunc['db_fetch_assoc']($request))
  572. $stop_words[] = $row['id_word'];
  573. $smcFunc['db_free_result']($request);
  574. updateSettings(array('search_stopwords' => implode(',', $stop_words)));
  575. if (!empty($stop_words))
  576. $smcFunc['db_query']('', '
  577. DELETE FROM {db_prefix}log_search_words
  578. WHERE id_word in ({array_int:stop_words})',
  579. array(
  580. 'stop_words' => $stop_words,
  581. )
  582. );
  583. $context['start'] += $index_properties[$context['index_settings']['bytes_per_word']]['step_size'];
  584. if ($context['start'] > $index_properties[$context['index_settings']['bytes_per_word']]['max_size'])
  585. {
  586. $context['step'] = 3;
  587. break;
  588. }
  589. }
  590. $context['percentage'] = 80 + round($context['start'] / $index_properties[$context['index_settings']['bytes_per_word']]['max_size'], 3) * 20;
  591. }
  592. }
  593. // Step 3: remove words not distinctive enough.
  594. if ($context['step'] === 3)
  595. {
  596. $context['sub_template'] = 'create_index_done';
  597. updateSettings(array('search_index' => 'custom', 'search_custom_index_config' => serialize($context['index_settings'])));
  598. $smcFunc['db_query']('', '
  599. DELETE FROM {db_prefix}settings
  600. WHERE variable = {string:search_custom_index_resume}',
  601. array(
  602. 'search_custom_index_resume' => 'search_custom_index_resume',
  603. )
  604. );
  605. }
  606. }
  607. /**
  608. * Get the installed Search API implementations.
  609. * This function checks for patterns in comments on top of the Search-API files!
  610. * In addition to filenames pattern.
  611. * It loads the search API classes if identified.
  612. * This function is used by EditSearchMethod to list all installed API implementations.
  613. */
  614. function loadSearchAPIs()
  615. {
  616. global $sourcedir, $txt;
  617. $apis = array();
  618. if ($dh = opendir($sourcedir))
  619. {
  620. while (($file = readdir($dh)) !== false)
  621. {
  622. if (is_file($sourcedir . '/' . $file) && preg_match('~^SearchAPI-([A-Za-z\d_]+)\.php$~', $file, $matches))
  623. {
  624. // Check this is definitely a valid API!
  625. $fp = fopen($sourcedir . '/' . $file, 'rb');
  626. $header = fread($fp, 4096);
  627. fclose($fp);
  628. if (strpos($header, '* SearchAPI-' . $matches[1] . '.php') !== false)
  629. {
  630. require_once($sourcedir . '/' . $file);
  631. $index_name = strtolower($matches[1]);
  632. $search_class_name = $index_name . '_search';
  633. $searchAPI = new $search_class_name();
  634. // No Support? NEXT!
  635. if (!$searchAPI->is_supported)
  636. continue;
  637. $apis[$index_name] = array(
  638. 'filename' => $file,
  639. 'setting_index' => $index_name,
  640. 'has_template' => in_array($index_name, array('custom', 'fulltext', 'standard')),
  641. 'label' => $index_name && isset($txt['search_index_' . $index_name]) ? $txt['search_index_' . $index_name] : '',
  642. 'desc' => $index_name && isset($txt['search_index_' . $index_name . '_desc']) ? $txt['search_index_' . $index_name . '_desc'] : '',
  643. );
  644. }
  645. }
  646. }
  647. }
  648. closedir($dh);
  649. return $apis;
  650. }
  651. /**
  652. * Checks if the message table already has a fulltext index created and returns the key name
  653. * Determines if a db is capable of creating a fulltext index
  654. */
  655. function detectFulltextIndex()
  656. {
  657. global $smcFunc, $context, $db_prefix;
  658. $request = $smcFunc['db_query']('', '
  659. SHOW INDEX
  660. FROM {db_prefix}messages',
  661. array(
  662. )
  663. );
  664. $context['fulltext_index'] = '';
  665. if ($request !== false || $smcFunc['db_num_rows']($request) != 0)
  666. {
  667. while ($row = $smcFunc['db_fetch_assoc']($request))
  668. if ($row['Column_name'] == 'body' && (isset($row['Index_type']) && $row['Index_type'] == 'FULLTEXT' || isset($row['Comment']) && $row['Comment'] == 'FULLTEXT'))
  669. $context['fulltext_index'][] = $row['Key_name'];
  670. $smcFunc['db_free_result']($request);
  671. if (is_array($context['fulltext_index']))
  672. $context['fulltext_index'] = array_unique($context['fulltext_index']);
  673. }
  674. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  675. $request = $smcFunc['db_query']('', '
  676. SHOW TABLE STATUS
  677. FROM {string:database_name}
  678. LIKE {string:table_name}',
  679. array(
  680. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  681. 'table_name' => str_replace('_', '\_', $match[2]) . 'messages',
  682. )
  683. );
  684. else
  685. $request = $smcFunc['db_query']('', '
  686. SHOW TABLE STATUS
  687. LIKE {string:table_name}',
  688. array(
  689. 'table_name' => str_replace('_', '\_', $db_prefix) . 'messages',
  690. )
  691. );
  692. if ($request !== false)
  693. {
  694. while ($row = $smcFunc['db_fetch_assoc']($request))
  695. if ((isset($row['Type']) && strtolower($row['Type']) != 'myisam') || (isset($row['Engine']) && strtolower($row['Engine']) != 'myisam'))
  696. $context['cannot_create_fulltext'] = true;
  697. $smcFunc['db_free_result']($request);
  698. }
  699. }
  700. ?>