ManageSearch.php 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798
  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 2011 Simple Machines
  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('Hacking attempt...');
  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-acton.
  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('int', 'search_results_per_page'),
  85. array('int', 'search_max_results', 'subtext' => $txt['search_max_results_disable']),
  86. '',
  87. // Some limitations.
  88. array('int', 'search_floodcontrol_time', 'subtext' => $txt['search_floodcontrol_time_desc']),
  89. );
  90. call_integration_hook('integrate_modify_search_settings', array(&$config_vars));
  91. // Perhaps the search method wants to add some settings?
  92. require_once($sourcedir . '/Search.php');
  93. $searchAPI = findSearchAPI();
  94. if (is_callable(array($searchAPI, 'searchSettings')))
  95. call_user_func_array($searchAPI->searchSettings, array(&$config_vars));
  96. if ($return_config)
  97. return $config_vars;
  98. $context['page_title'] = $txt['search_settings_title'];
  99. $context['sub_template'] = 'show_settings';
  100. call_integration_hook('integrate_modify_search_weights', array(&$factors));
  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'] = $txt['search_settings_title'];
  114. prepareDBSettingContext($config_vars);
  115. }
  116. /**
  117. * Edit the relative weight of the search factors.
  118. * Called by ?action=admin;area=managesearch;sa=weights.
  119. * Requires the admin_forum permission.
  120. *
  121. * @uses ManageSearch template, 'modify_weights' sub-template.
  122. */
  123. function EditWeights()
  124. {
  125. global $txt, $context, $modSettings;
  126. $context['page_title'] = $txt['search_weights_title'];
  127. $context['sub_template'] = 'modify_weights';
  128. $factors = array(
  129. 'search_weight_frequency',
  130. 'search_weight_age',
  131. 'search_weight_length',
  132. 'search_weight_subject',
  133. 'search_weight_first_message',
  134. 'search_weight_sticky',
  135. );
  136. call_integration_hook('integrate_modify_search_weights', array(&$factors));
  137. // A form was submitted.
  138. if (isset($_POST['save']))
  139. {
  140. checkSession();
  141. validateToken('admin-msw');
  142. call_integration_hook('integrate_save_search_weights');
  143. $changes = array();
  144. foreach ($factors as $factor)
  145. $changes[$factor] = (int) $_POST[$factor];
  146. updateSettings($changes);
  147. }
  148. $context['relative_weights'] = array('total' => 0);
  149. foreach ($factors as $factor)
  150. $context['relative_weights']['total'] += isset($modSettings[$factor]) ? $modSettings[$factor] : 0;
  151. foreach ($factors as $factor)
  152. $context['relative_weights'][$factor] = round(100 * (isset($modSettings[$factor]) ? $modSettings[$factor] : 0) / $context['relative_weights']['total'], 1);
  153. createToken('admin-msw');
  154. }
  155. /**
  156. * Edit the search method and search index used.
  157. * Calculates the size of the current search indexes in use.
  158. * Allows to create and delete a fulltext index on the messages table.
  159. * Allows to delete a custom index (that CreateMessageIndex() created).
  160. * Called by ?action=admin;area=managesearch;sa=method.
  161. * Requires the admin_forum permission.
  162. *
  163. * @uses ManageSearch template, 'select_search_method' sub-template.
  164. */
  165. function EditSearchMethod()
  166. {
  167. global $txt, $context, $modSettings, $smcFunc, $db_type, $db_prefix;
  168. $context[$context['admin_menu_name']]['current_subsection'] = 'method';
  169. $context['page_title'] = $txt['search_method_title'];
  170. $context['sub_template'] = 'select_search_method';
  171. $context['supports_fulltext'] = $smcFunc['db_search_support']('fulltext');
  172. // Load any apis.
  173. $context['search_apis'] = loadSearchAPIs();
  174. // Detect whether a fulltext index is set.
  175. if ($context['supports_fulltext'])
  176. {
  177. $request = $smcFunc['db_query']('', '
  178. SHOW INDEX
  179. FROM {db_prefix}messages',
  180. array(
  181. )
  182. );
  183. $context['fulltext_index'] = '';
  184. if ($request !== false || $smcFunc['db_num_rows']($request) != 0)
  185. {
  186. while ($row = $smcFunc['db_fetch_assoc']($request))
  187. if ($row['Column_name'] == 'body' && (isset($row['Index_type']) && $row['Index_type'] == 'FULLTEXT' || isset($row['Comment']) && $row['Comment'] == 'FULLTEXT'))
  188. $context['fulltext_index'][] = $row['Key_name'];
  189. $smcFunc['db_free_result']($request);
  190. if (is_array($context['fulltext_index']))
  191. $context['fulltext_index'] = array_unique($context['fulltext_index']);
  192. }
  193. $request = $smcFunc['db_query']('', '
  194. SHOW COLUMNS
  195. FROM {db_prefix}messages',
  196. array(
  197. )
  198. );
  199. if ($request !== false)
  200. {
  201. while ($row = $smcFunc['db_fetch_assoc']($request))
  202. if ($row['Field'] == 'body' && $row['Type'] == 'mediumtext')
  203. $context['cannot_create_fulltext'] = true;
  204. $smcFunc['db_free_result']($request);
  205. }
  206. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  207. $request = $smcFunc['db_query']('', '
  208. SHOW TABLE STATUS
  209. FROM {string:database_name}
  210. LIKE {string:table_name}',
  211. array(
  212. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  213. 'table_name' => str_replace('_', '\_', $match[2]) . 'messages',
  214. )
  215. );
  216. else
  217. $request = $smcFunc['db_query']('', '
  218. SHOW TABLE STATUS
  219. LIKE {string:table_name}',
  220. array(
  221. 'table_name' => str_replace('_', '\_', $db_prefix) . 'messages',
  222. )
  223. );
  224. if ($request !== false)
  225. {
  226. while ($row = $smcFunc['db_fetch_assoc']($request))
  227. if ((isset($row['Type']) && strtolower($row['Type']) != 'myisam') || (isset($row['Engine']) && strtolower($row['Engine']) != 'myisam'))
  228. $context['cannot_create_fulltext'] = true;
  229. $smcFunc['db_free_result']($request);
  230. }
  231. }
  232. if (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'createfulltext')
  233. {
  234. checkSession('get');
  235. validateToken('admin-msm');
  236. // Make sure it's gone before creating it.
  237. $smcFunc['db_query']('', '
  238. ALTER TABLE {db_prefix}messages
  239. DROP INDEX body',
  240. array(
  241. 'db_error_skip' => true,
  242. )
  243. );
  244. $smcFunc['db_query']('', '
  245. ALTER TABLE {db_prefix}messages
  246. ADD FULLTEXT body (body)',
  247. array(
  248. )
  249. );
  250. $context['fulltext_index'] = 'body';
  251. }
  252. elseif (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'removefulltext' && !empty($context['fulltext_index']))
  253. {
  254. checkSession('get');
  255. validateToken('admin-msm');
  256. $smcFunc['db_query']('', '
  257. ALTER TABLE {db_prefix}messages
  258. DROP INDEX ' . implode(',
  259. DROP INDEX ', $context['fulltext_index']),
  260. array(
  261. 'db_error_skip' => true,
  262. )
  263. );
  264. $context['fulltext_index'] = '';
  265. // Go back to the default search method.
  266. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'fulltext')
  267. updateSettings(array(
  268. 'search_index' => '',
  269. ));
  270. }
  271. elseif (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'removecustom')
  272. {
  273. checkSession('get');
  274. validateToken('admin-msm');
  275. db_extend();
  276. $tables = $smcFunc['db_list_tables'](false, $db_prefix . 'log_search_words');
  277. if (!empty($tables))
  278. {
  279. $smcFunc['db_search_query']('drop_words_table', '
  280. DROP TABLE {db_prefix}log_search_words',
  281. array(
  282. )
  283. );
  284. }
  285. updateSettings(array(
  286. 'search_custom_index_config' => '',
  287. 'search_custom_index_resume' => '',
  288. ));
  289. // Go back to the default search method.
  290. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'custom')
  291. updateSettings(array(
  292. 'search_index' => '',
  293. ));
  294. }
  295. elseif (isset($_POST['save']))
  296. {
  297. checkSession();
  298. validateToken('admin-msm');
  299. updateSettings(array(
  300. 'search_index' => empty($_POST['search_index']) || (!in_array($_POST['search_index'], array('fulltext', 'custom')) && !isset($context['search_apis'][$_POST['search_index']])) ? '' : $_POST['search_index'],
  301. 'search_force_index' => isset($_POST['search_force_index']) ? '1' : '0',
  302. 'search_match_words' => isset($_POST['search_match_words']) ? '1' : '0',
  303. ));
  304. }
  305. $context['table_info'] = array(
  306. 'data_length' => 0,
  307. 'index_length' => 0,
  308. 'fulltext_length' => 0,
  309. 'custom_index_length' => 0,
  310. );
  311. // Get some info about the messages table, to show its size and index size.
  312. if ($db_type == 'mysql')
  313. {
  314. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  315. $request = $smcFunc['db_query']('', '
  316. SHOW TABLE STATUS
  317. FROM {string:database_name}
  318. LIKE {string:table_name}',
  319. array(
  320. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  321. 'table_name' => str_replace('_', '\_', $match[2]) . 'messages',
  322. )
  323. );
  324. else
  325. $request = $smcFunc['db_query']('', '
  326. SHOW TABLE STATUS
  327. LIKE {string:table_name}',
  328. array(
  329. 'table_name' => str_replace('_', '\_', $db_prefix) . 'messages',
  330. )
  331. );
  332. if ($request !== false && $smcFunc['db_num_rows']($request) == 1)
  333. {
  334. // Only do this if the user has permission to execute this query.
  335. $row = $smcFunc['db_fetch_assoc']($request);
  336. $context['table_info']['data_length'] = $row['Data_length'];
  337. $context['table_info']['index_length'] = $row['Index_length'];
  338. $context['table_info']['fulltext_length'] = $row['Index_length'];
  339. $smcFunc['db_free_result']($request);
  340. }
  341. // Now check the custom index table, if it exists at all.
  342. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  343. $request = $smcFunc['db_query']('', '
  344. SHOW TABLE STATUS
  345. FROM {string:database_name}
  346. LIKE {string:table_name}',
  347. array(
  348. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  349. 'table_name' => str_replace('_', '\_', $match[2]) . 'log_search_words',
  350. )
  351. );
  352. else
  353. $request = $smcFunc['db_query']('', '
  354. SHOW TABLE STATUS
  355. LIKE {string:table_name}',
  356. array(
  357. 'table_name' => str_replace('_', '\_', $db_prefix) . 'log_search_words',
  358. )
  359. );
  360. if ($request !== false && $smcFunc['db_num_rows']($request) == 1)
  361. {
  362. // Only do this if the user has permission to execute this query.
  363. $row = $smcFunc['db_fetch_assoc']($request);
  364. $context['table_info']['index_length'] += $row['Data_length'] + $row['Index_length'];
  365. $context['table_info']['custom_index_length'] = $row['Data_length'] + $row['Index_length'];
  366. $smcFunc['db_free_result']($request);
  367. }
  368. }
  369. elseif ($db_type == 'postgresql')
  370. {
  371. // In order to report the sizes correctly we need to perform vacuum (optimize) on the tables we will be using.
  372. db_extend();
  373. $temp_tables = $smcFunc['db_list_tables']();
  374. foreach ($temp_tables as $table)
  375. if ($table == $db_prefix. 'messages' || $table == $db_prefix. 'log_search_words')
  376. $smcFunc['db_optimize_table']($table);
  377. // PostGreSql has some hidden sizes.
  378. $request = $smcFunc['db_query']('', '
  379. SELECT relname, relpages * 8 *1024 AS "KB" FROM pg_class
  380. WHERE relname = {string:messages} OR relname = {string:log_search_words}
  381. ORDER BY relpages DESC',
  382. array(
  383. 'messages' => $db_prefix. 'messages',
  384. 'log_search_words' => $db_prefix. 'log_search_words',
  385. )
  386. );
  387. if ($request !== false && $smcFunc['db_num_rows']($request) > 0)
  388. {
  389. while ($row = $smcFunc['db_fetch_assoc']($request))
  390. {
  391. if ($row['relname'] == $db_prefix . 'messages')
  392. {
  393. $context['table_info']['data_length'] = (int) $row['KB'];
  394. $context['table_info']['index_length'] = (int) $row['KB'];
  395. // Doesn't support fulltext
  396. $context['table_info']['fulltext_length'] = $txt['not_applicable'];
  397. }
  398. elseif ($row['relname'] == $db_prefix. 'log_search_words')
  399. {
  400. $context['table_info']['index_length'] = (int) $row['KB'];
  401. $context['table_info']['custom_index_length'] = (int) $row['KB'];
  402. }
  403. }
  404. $smcFunc['db_free_result']($request);
  405. }
  406. else
  407. // Didn't work for some reason...
  408. $context['table_info'] = array(
  409. 'data_length' => $txt['not_applicable'],
  410. 'index_length' => $txt['not_applicable'],
  411. 'fulltext_length' => $txt['not_applicable'],
  412. 'custom_index_length' => $txt['not_applicable'],
  413. );
  414. }
  415. else
  416. $context['table_info'] = array(
  417. 'data_length' => $txt['not_applicable'],
  418. 'index_length' => $txt['not_applicable'],
  419. 'fulltext_length' => $txt['not_applicable'],
  420. 'custom_index_length' => $txt['not_applicable'],
  421. );
  422. // Format the data and index length in kilobytes.
  423. foreach ($context['table_info'] as $type => $size)
  424. {
  425. // If it's not numeric then just break. This database engine doesn't support size.
  426. if (!is_numeric($size))
  427. break;
  428. $context['table_info'][$type] = comma_format($context['table_info'][$type] / 1024) . ' ' . $txt['search_method_kilobytes'];
  429. }
  430. $context['custom_index'] = !empty($modSettings['search_custom_index_config']);
  431. $context['partial_custom_index'] = !empty($modSettings['search_custom_index_resume']) && empty($modSettings['search_custom_index_config']);
  432. $context['double_index'] = !empty($context['fulltext_index']) && $context['custom_index'];
  433. createToken('admin-msm');
  434. }
  435. /**
  436. * Create a custom search index for the messages table.
  437. * Called by ?action=admin;area=managesearch;sa=createmsgindex.
  438. * Linked from the EditSearchMethod screen.
  439. * Requires the admin_forum permission.
  440. * Depending on the size of the message table, the process is divided in steps.
  441. *
  442. * @uses ManageSearch template, 'create_index', 'create_index_progress', and 'create_index_done'
  443. * sub-templates.
  444. */
  445. function CreateMessageIndex()
  446. {
  447. global $modSettings, $context, $smcFunc, $db_prefix, $txt;
  448. // Scotty, we need more time...
  449. @set_time_limit(600);
  450. if (function_exists('apache_reset_timeout'))
  451. @apache_reset_timeout();
  452. $context[$context['admin_menu_name']]['current_subsection'] = 'method';
  453. $context['page_title'] = $txt['search_index_custom'];
  454. $messages_per_batch = 50;
  455. $index_properties = array(
  456. 2 => array(
  457. 'column_definition' => 'small',
  458. 'step_size' => 1000000,
  459. ),
  460. 4 => array(
  461. 'column_definition' => 'medium',
  462. 'step_size' => 1000000,
  463. 'max_size' => 16777215,
  464. ),
  465. 5 => array(
  466. 'column_definition' => 'large',
  467. 'step_size' => 100000000,
  468. 'max_size' => 2000000000,
  469. ),
  470. );
  471. if (isset($_REQUEST['resume']) && !empty($modSettings['search_custom_index_resume']))
  472. {
  473. $context['index_settings'] = unserialize($modSettings['search_custom_index_resume']);
  474. $context['start'] = (int) $context['index_settings']['resume_at'];
  475. unset($context['index_settings']['resume_at']);
  476. $context['step'] = 1;
  477. }
  478. else
  479. {
  480. $context['index_settings'] = array(
  481. 'bytes_per_word' => isset($_REQUEST['bytes_per_word']) && isset($index_properties[$_REQUEST['bytes_per_word']]) ? (int) $_REQUEST['bytes_per_word'] : 2,
  482. );
  483. $context['start'] = isset($_REQUEST['start']) ? (int) $_REQUEST['start'] : 0;
  484. $context['step'] = isset($_REQUEST['step']) ? (int) $_REQUEST['step'] : 0;
  485. }
  486. if ($context['step'] !== 0)
  487. checkSession('request');
  488. // Step 0: let the user determine how they like their index.
  489. if ($context['step'] === 0)
  490. {
  491. $context['sub_template'] = 'create_index';
  492. }
  493. // Step 1: insert all the words.
  494. if ($context['step'] === 1)
  495. {
  496. $context['sub_template'] = 'create_index_progress';
  497. if ($context['start'] === 0)
  498. {
  499. db_extend();
  500. $tables = $smcFunc['db_list_tables'](false, $db_prefix . 'log_search_words');
  501. if (!empty($tables))
  502. {
  503. $smcFunc['db_search_query']('drop_words_table', '
  504. DROP TABLE {db_prefix}log_search_words',
  505. array(
  506. )
  507. );
  508. }
  509. $smcFunc['db_create_word_search']($index_properties[$context['index_settings']['bytes_per_word']]['column_definition']);
  510. // Temporarily switch back to not using a search index.
  511. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'custom')
  512. updateSettings(array('search_index' => ''));
  513. // Don't let simultanious processes be updating the search index.
  514. if (!empty($modSettings['search_custom_index_config']))
  515. updateSettings(array('search_custom_index_config' => ''));
  516. }
  517. $num_messages = array(
  518. 'done' => 0,
  519. 'todo' => 0,
  520. );
  521. $request = $smcFunc['db_query']('', '
  522. SELECT id_msg >= {int:starting_id} AS todo, COUNT(*) AS num_messages
  523. FROM {db_prefix}messages
  524. GROUP BY todo',
  525. array(
  526. 'starting_id' => $context['start'],
  527. )
  528. );
  529. while ($row = $smcFunc['db_fetch_assoc']($request))
  530. $num_messages[empty($row['todo']) ? 'done' : 'todo'] = $row['num_messages'];
  531. if (empty($num_messages['todo']))
  532. {
  533. $context['step'] = 2;
  534. $context['percentage'] = 80;
  535. $context['start'] = 0;
  536. }
  537. else
  538. {
  539. // Number of seconds before the next step.
  540. $stop = time() + 3;
  541. while (time() < $stop)
  542. {
  543. $inserts = array();
  544. $request = $smcFunc['db_query']('', '
  545. SELECT id_msg, body
  546. FROM {db_prefix}messages
  547. WHERE id_msg BETWEEN {int:starting_id} AND {int:ending_id}
  548. LIMIT {int:limit}',
  549. array(
  550. 'starting_id' => $context['start'],
  551. 'ending_id' => $context['start'] + $messages_per_batch - 1,
  552. 'limit' => $messages_per_batch,
  553. )
  554. );
  555. $forced_break = false;
  556. $number_processed = 0;
  557. while ($row = $smcFunc['db_fetch_assoc']($request))
  558. {
  559. // In theory it's possible for one of these to take friggin ages so add more timeout protection.
  560. if ($stop < time())
  561. {
  562. $forced_break = true;
  563. break;
  564. }
  565. $number_processed++;
  566. foreach (text2words($row['body'], $context['index_settings']['bytes_per_word'], true) as $id_word)
  567. {
  568. $inserts[] = array($id_word, $row['id_msg']);
  569. }
  570. }
  571. $num_messages['done'] += $number_processed;
  572. $num_messages['todo'] -= $number_processed;
  573. $smcFunc['db_free_result']($request);
  574. $context['start'] += $forced_break ? $number_processed : $messages_per_batch;
  575. if (!empty($inserts))
  576. $smcFunc['db_insert']('ignore',
  577. '{db_prefix}log_search_words',
  578. array('id_word' => 'int', 'id_msg' => 'int'),
  579. $inserts,
  580. array('id_word', 'id_msg')
  581. );
  582. if ($num_messages['todo'] === 0)
  583. {
  584. $context['step'] = 2;
  585. $context['start'] = 0;
  586. break;
  587. }
  588. else
  589. updateSettings(array('search_custom_index_resume' => serialize(array_merge($context['index_settings'], array('resume_at' => $context['start'])))));
  590. }
  591. // Since there are still two steps to go, 90% is the maximum here.
  592. $context['percentage'] = round($num_messages['done'] / ($num_messages['done'] + $num_messages['todo']), 3) * 80;
  593. }
  594. }
  595. // Step 2: removing the words that occur too often and are of no use.
  596. elseif ($context['step'] === 2)
  597. {
  598. if ($context['index_settings']['bytes_per_word'] < 4)
  599. $context['step'] = 3;
  600. else
  601. {
  602. $stop_words = $context['start'] === 0 || empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
  603. $stop = time() + 3;
  604. $context['sub_template'] = 'create_index_progress';
  605. $max_messages = ceil(60 * $modSettings['totalMessages'] / 100);
  606. while (time() < $stop)
  607. {
  608. $request = $smcFunc['db_query']('', '
  609. SELECT id_word, COUNT(id_word) AS num_words
  610. FROM {db_prefix}log_search_words
  611. WHERE id_word BETWEEN {int:starting_id} AND {int:ending_id}
  612. GROUP BY id_word
  613. HAVING COUNT(id_word) > {int:minimum_messages}',
  614. array(
  615. 'starting_id' => $context['start'],
  616. 'ending_id' => $context['start'] + $index_properties[$context['index_settings']['bytes_per_word']]['step_size'] - 1,
  617. 'minimum_messages' => $max_messages,
  618. )
  619. );
  620. while ($row = $smcFunc['db_fetch_assoc']($request))
  621. $stop_words[] = $row['id_word'];
  622. $smcFunc['db_free_result']($request);
  623. updateSettings(array('search_stopwords' => implode(',', $stop_words)));
  624. if (!empty($stop_words))
  625. $smcFunc['db_query']('', '
  626. DELETE FROM {db_prefix}log_search_words
  627. WHERE id_word in ({array_int:stop_words})',
  628. array(
  629. 'stop_words' => $stop_words,
  630. )
  631. );
  632. $context['start'] += $index_properties[$context['index_settings']['bytes_per_word']]['step_size'];
  633. if ($context['start'] > $index_properties[$context['index_settings']['bytes_per_word']]['max_size'])
  634. {
  635. $context['step'] = 3;
  636. break;
  637. }
  638. }
  639. $context['percentage'] = 80 + round($context['start'] / $index_properties[$context['index_settings']['bytes_per_word']]['max_size'], 3) * 20;
  640. }
  641. }
  642. // Step 3: remove words not distinctive enough.
  643. if ($context['step'] === 3)
  644. {
  645. $context['sub_template'] = 'create_index_done';
  646. updateSettings(array('search_index' => 'custom', 'search_custom_index_config' => serialize($context['index_settings'])));
  647. $smcFunc['db_query']('', '
  648. DELETE FROM {db_prefix}settings
  649. WHERE variable = {string:search_custom_index_resume}',
  650. array(
  651. 'search_custom_index_resume' => 'search_custom_index_resume',
  652. )
  653. );
  654. }
  655. }
  656. /**
  657. * Get the installed Search API implementations.
  658. * This function checks for patterns in comments on top of the Search-API files!
  659. * In addition to filenames pattern.
  660. * It loads the search API classes if identified.
  661. * This function is used by EditSearchMethod to list all installed API implementations.
  662. */
  663. function loadSearchAPIs()
  664. {
  665. global $sourcedir, $txt;
  666. $apis = array();
  667. if ($dh = opendir($sourcedir))
  668. {
  669. while (($file = readdir($dh)) !== false)
  670. {
  671. if (is_file($sourcedir . '/' . $file) && preg_match('~^SearchAPI-([A-Za-z\d_]+)\.php$~', $file, $matches))
  672. {
  673. // Check this is definitely a valid API!
  674. $fp = fopen($sourcedir . '/' . $file, 'rb');
  675. $header = fread($fp, 4096);
  676. fclose($fp);
  677. if (strpos($header, '* SearchAPI-' . $matches[1] . '.php') !== false)
  678. {
  679. require_once($sourcedir . '/' . $file);
  680. $index_name = strtolower($matches[1]);
  681. $search_class_name = $index_name . '_search';
  682. $searchAPI = new $search_class_name();
  683. // No Support? NEXT!
  684. if (!$searchAPI->is_supported)
  685. continue;
  686. $apis[$index_name] = array(
  687. 'filename' => $file,
  688. 'setting_index' => $index_name,
  689. 'has_template' => in_array($index_name, array('custom', 'fulltext', 'standard')),
  690. 'label' => $index_name && isset($txt['search_index_' . $index_name]) ? $txt['search_index_' . $index_name] : '',
  691. 'desc' => $index_name && isset($txt['search_index_' . $index_name . '_desc']) ? $txt['search_index_' . $index_name . '_desc'] : '',
  692. );
  693. }
  694. }
  695. }
  696. }
  697. closedir($dh);
  698. return $apis;
  699. }
  700. ?>