ManageSearch.php 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794
  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. $factors = array(
  129. 'search_weight_frequency',
  130. 'search_weight_age',
  131. 'search_weight_length',
  132. 'search_weight_subject',
  133. 'search_weight_first_message',
  134. 'search_weight_sticky',
  135. );
  136. // A form was submitted.
  137. if (isset($_POST['save']))
  138. {
  139. checkSession();
  140. validateToken('admin-msw');
  141. $changes = array();
  142. foreach ($factors as $factor)
  143. $changes[$factor] = (int) $_POST[$factor];
  144. updateSettings($changes);
  145. }
  146. $context['relative_weights'] = array('total' => 0);
  147. foreach ($factors as $factor)
  148. $context['relative_weights']['total'] += isset($modSettings[$factor]) ? $modSettings[$factor] : 0;
  149. foreach ($factors as $factor)
  150. $context['relative_weights'][$factor] = round(100 * (isset($modSettings[$factor]) ? $modSettings[$factor] : 0) / $context['relative_weights']['total'], 1);
  151. createToken('admin-msw');
  152. }
  153. /**
  154. * Edit the search method and search index used.
  155. * Calculates the size of the current search indexes in use.
  156. * Allows to create and delete a fulltext index on the messages table.
  157. * Allows to delete a custom index (that CreateMessageIndex() created).
  158. * Called by ?action=admin;area=managesearch;sa=method.
  159. * Requires the admin_forum permission.
  160. *
  161. * @uses ManageSearch template, 'select_search_method' sub-template.
  162. */
  163. function EditSearchMethod()
  164. {
  165. global $txt, $context, $modSettings, $smcFunc, $db_type, $db_prefix;
  166. $context[$context['admin_menu_name']]['current_subsection'] = 'method';
  167. $context['page_title'] = $txt['search_method_title'];
  168. $context['sub_template'] = 'select_search_method';
  169. $context['supports_fulltext'] = $smcFunc['db_search_support']('fulltext');
  170. // Load any apis.
  171. $context['search_apis'] = loadSearchAPIs();
  172. // Detect whether a fulltext index is set.
  173. if ($context['supports_fulltext'])
  174. {
  175. $request = $smcFunc['db_query']('', '
  176. SHOW INDEX
  177. FROM {db_prefix}messages',
  178. array(
  179. )
  180. );
  181. $context['fulltext_index'] = '';
  182. if ($request !== false || $smcFunc['db_num_rows']($request) != 0)
  183. {
  184. while ($row = $smcFunc['db_fetch_assoc']($request))
  185. if ($row['Column_name'] == 'body' && (isset($row['Index_type']) && $row['Index_type'] == 'FULLTEXT' || isset($row['Comment']) && $row['Comment'] == 'FULLTEXT'))
  186. $context['fulltext_index'][] = $row['Key_name'];
  187. $smcFunc['db_free_result']($request);
  188. if (is_array($context['fulltext_index']))
  189. $context['fulltext_index'] = array_unique($context['fulltext_index']);
  190. }
  191. $request = $smcFunc['db_query']('', '
  192. SHOW COLUMNS
  193. FROM {db_prefix}messages',
  194. array(
  195. )
  196. );
  197. if ($request !== false)
  198. {
  199. while ($row = $smcFunc['db_fetch_assoc']($request))
  200. if ($row['Field'] == 'body' && $row['Type'] == 'mediumtext')
  201. $context['cannot_create_fulltext'] = true;
  202. $smcFunc['db_free_result']($request);
  203. }
  204. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  205. $request = $smcFunc['db_query']('', '
  206. SHOW TABLE STATUS
  207. FROM {string:database_name}
  208. LIKE {string:table_name}',
  209. array(
  210. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  211. 'table_name' => str_replace('_', '\_', $match[2]) . 'messages',
  212. )
  213. );
  214. else
  215. $request = $smcFunc['db_query']('', '
  216. SHOW TABLE STATUS
  217. LIKE {string:table_name}',
  218. array(
  219. 'table_name' => str_replace('_', '\_', $db_prefix) . 'messages',
  220. )
  221. );
  222. if ($request !== false)
  223. {
  224. while ($row = $smcFunc['db_fetch_assoc']($request))
  225. if ((isset($row['Type']) && strtolower($row['Type']) != 'myisam') || (isset($row['Engine']) && strtolower($row['Engine']) != 'myisam'))
  226. $context['cannot_create_fulltext'] = true;
  227. $smcFunc['db_free_result']($request);
  228. }
  229. }
  230. if (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'createfulltext')
  231. {
  232. checkSession('get');
  233. validateToken('admin-msm');
  234. // Make sure it's gone before creating it.
  235. $smcFunc['db_query']('', '
  236. ALTER TABLE {db_prefix}messages
  237. DROP INDEX body',
  238. array(
  239. 'db_error_skip' => true,
  240. )
  241. );
  242. $smcFunc['db_query']('', '
  243. ALTER TABLE {db_prefix}messages
  244. ADD FULLTEXT body (body)',
  245. array(
  246. )
  247. );
  248. $context['fulltext_index'] = 'body';
  249. }
  250. elseif (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'removefulltext' && !empty($context['fulltext_index']))
  251. {
  252. checkSession('get');
  253. validateToken('admin-msm');
  254. $smcFunc['db_query']('', '
  255. ALTER TABLE {db_prefix}messages
  256. DROP INDEX ' . implode(',
  257. DROP INDEX ', $context['fulltext_index']),
  258. array(
  259. 'db_error_skip' => true,
  260. )
  261. );
  262. $context['fulltext_index'] = '';
  263. // Go back to the default search method.
  264. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'fulltext')
  265. updateSettings(array(
  266. 'search_index' => '',
  267. ));
  268. }
  269. elseif (!empty($_REQUEST['sa']) && $_REQUEST['sa'] == 'removecustom')
  270. {
  271. checkSession('get');
  272. validateToken('admin-msm');
  273. db_extend();
  274. $tables = $smcFunc['db_list_tables'](false, $db_prefix . 'log_search_words');
  275. if (!empty($tables))
  276. {
  277. $smcFunc['db_search_query']('drop_words_table', '
  278. DROP TABLE {db_prefix}log_search_words',
  279. array(
  280. )
  281. );
  282. }
  283. updateSettings(array(
  284. 'search_custom_index_config' => '',
  285. 'search_custom_index_resume' => '',
  286. ));
  287. // Go back to the default search method.
  288. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'custom')
  289. updateSettings(array(
  290. 'search_index' => '',
  291. ));
  292. }
  293. elseif (isset($_POST['save']))
  294. {
  295. checkSession();
  296. validateToken('admin-msm');
  297. updateSettings(array(
  298. 'search_index' => empty($_POST['search_index']) || (!in_array($_POST['search_index'], array('fulltext', 'custom')) && !isset($context['search_apis'][$_POST['search_index']])) ? '' : $_POST['search_index'],
  299. 'search_force_index' => isset($_POST['search_force_index']) ? '1' : '0',
  300. 'search_match_words' => isset($_POST['search_match_words']) ? '1' : '0',
  301. ));
  302. }
  303. $context['table_info'] = array(
  304. 'data_length' => 0,
  305. 'index_length' => 0,
  306. 'fulltext_length' => 0,
  307. 'custom_index_length' => 0,
  308. );
  309. // Get some info about the messages table, to show its size and index size.
  310. if ($db_type == 'mysql')
  311. {
  312. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  313. $request = $smcFunc['db_query']('', '
  314. SHOW TABLE STATUS
  315. FROM {string:database_name}
  316. LIKE {string:table_name}',
  317. array(
  318. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  319. 'table_name' => str_replace('_', '\_', $match[2]) . 'messages',
  320. )
  321. );
  322. else
  323. $request = $smcFunc['db_query']('', '
  324. SHOW TABLE STATUS
  325. LIKE {string:table_name}',
  326. array(
  327. 'table_name' => str_replace('_', '\_', $db_prefix) . 'messages',
  328. )
  329. );
  330. if ($request !== false && $smcFunc['db_num_rows']($request) == 1)
  331. {
  332. // Only do this if the user has permission to execute this query.
  333. $row = $smcFunc['db_fetch_assoc']($request);
  334. $context['table_info']['data_length'] = $row['Data_length'];
  335. $context['table_info']['index_length'] = $row['Index_length'];
  336. $context['table_info']['fulltext_length'] = $row['Index_length'];
  337. $smcFunc['db_free_result']($request);
  338. }
  339. // Now check the custom index table, if it exists at all.
  340. if (preg_match('~^`(.+?)`\.(.+?)$~', $db_prefix, $match) !== 0)
  341. $request = $smcFunc['db_query']('', '
  342. SHOW TABLE STATUS
  343. FROM {string:database_name}
  344. LIKE {string:table_name}',
  345. array(
  346. 'database_name' => '`' . strtr($match[1], array('`' => '')) . '`',
  347. 'table_name' => str_replace('_', '\_', $match[2]) . 'log_search_words',
  348. )
  349. );
  350. else
  351. $request = $smcFunc['db_query']('', '
  352. SHOW TABLE STATUS
  353. LIKE {string:table_name}',
  354. array(
  355. 'table_name' => str_replace('_', '\_', $db_prefix) . 'log_search_words',
  356. )
  357. );
  358. if ($request !== false && $smcFunc['db_num_rows']($request) == 1)
  359. {
  360. // Only do this if the user has permission to execute this query.
  361. $row = $smcFunc['db_fetch_assoc']($request);
  362. $context['table_info']['index_length'] += $row['Data_length'] + $row['Index_length'];
  363. $context['table_info']['custom_index_length'] = $row['Data_length'] + $row['Index_length'];
  364. $smcFunc['db_free_result']($request);
  365. }
  366. }
  367. elseif ($db_type == 'postgresql')
  368. {
  369. // In order to report the sizes correctly we need to perform vacuum (optimize) on the tables we will be using.
  370. db_extend();
  371. $temp_tables = $smcFunc['db_list_tables']();
  372. foreach ($temp_tables as $table)
  373. if ($table == $db_prefix. 'messages' || $table == $db_prefix. 'log_search_words')
  374. $smcFunc['db_optimize_table']($table);
  375. // PostGreSql has some hidden sizes.
  376. $request = $smcFunc['db_query']('', '
  377. SELECT relname, relpages * 8 *1024 AS "KB" FROM pg_class
  378. WHERE relname = {string:messages} OR relname = {string:log_search_words}
  379. ORDER BY relpages DESC',
  380. array(
  381. 'messages' => $db_prefix. 'messages',
  382. 'log_search_words' => $db_prefix. 'log_search_words',
  383. )
  384. );
  385. if ($request !== false && $smcFunc['db_num_rows']($request) > 0)
  386. {
  387. while ($row = $smcFunc['db_fetch_assoc']($request))
  388. {
  389. if ($row['relname'] == $db_prefix . 'messages')
  390. {
  391. $context['table_info']['data_length'] = (int) $row['KB'];
  392. $context['table_info']['index_length'] = (int) $row['KB'];
  393. // Doesn't support fulltext
  394. $context['table_info']['fulltext_length'] = $txt['not_applicable'];
  395. }
  396. elseif ($row['relname'] == $db_prefix. 'log_search_words')
  397. {
  398. $context['table_info']['index_length'] = (int) $row['KB'];
  399. $context['table_info']['custom_index_length'] = (int) $row['KB'];
  400. }
  401. }
  402. $smcFunc['db_free_result']($request);
  403. }
  404. else
  405. // Didn't work for some reason...
  406. $context['table_info'] = array(
  407. 'data_length' => $txt['not_applicable'],
  408. 'index_length' => $txt['not_applicable'],
  409. 'fulltext_length' => $txt['not_applicable'],
  410. 'custom_index_length' => $txt['not_applicable'],
  411. );
  412. }
  413. else
  414. $context['table_info'] = array(
  415. 'data_length' => $txt['not_applicable'],
  416. 'index_length' => $txt['not_applicable'],
  417. 'fulltext_length' => $txt['not_applicable'],
  418. 'custom_index_length' => $txt['not_applicable'],
  419. );
  420. // Format the data and index length in kilobytes.
  421. foreach ($context['table_info'] as $type => $size)
  422. {
  423. // If it's not numeric then just break. This database engine doesn't support size.
  424. if (!is_numeric($size))
  425. break;
  426. $context['table_info'][$type] = comma_format($context['table_info'][$type] / 1024) . ' ' . $txt['search_method_kilobytes'];
  427. }
  428. $context['custom_index'] = !empty($modSettings['search_custom_index_config']);
  429. $context['partial_custom_index'] = !empty($modSettings['search_custom_index_resume']) && empty($modSettings['search_custom_index_config']);
  430. $context['double_index'] = !empty($context['fulltext_index']) && $context['custom_index'];
  431. createToken('admin-msm');
  432. }
  433. /**
  434. * Create a custom search index for the messages table.
  435. * Called by ?action=admin;area=managesearch;sa=createmsgindex.
  436. * Linked from the EditSearchMethod screen.
  437. * Requires the admin_forum permission.
  438. * Depending on the size of the message table, the process is divided in steps.
  439. *
  440. * @uses ManageSearch template, 'create_index', 'create_index_progress', and 'create_index_done'
  441. * sub-templates.
  442. */
  443. function CreateMessageIndex()
  444. {
  445. global $modSettings, $context, $smcFunc, $db_prefix, $txt;
  446. // Scotty, we need more time...
  447. @set_time_limit(600);
  448. if (function_exists('apache_reset_timeout'))
  449. @apache_reset_timeout();
  450. $context[$context['admin_menu_name']]['current_subsection'] = 'method';
  451. $context['page_title'] = $txt['search_index_custom'];
  452. $messages_per_batch = 50;
  453. $index_properties = array(
  454. 2 => array(
  455. 'column_definition' => 'small',
  456. 'step_size' => 1000000,
  457. ),
  458. 4 => array(
  459. 'column_definition' => 'medium',
  460. 'step_size' => 1000000,
  461. 'max_size' => 16777215,
  462. ),
  463. 5 => array(
  464. 'column_definition' => 'large',
  465. 'step_size' => 100000000,
  466. 'max_size' => 2000000000,
  467. ),
  468. );
  469. if (isset($_REQUEST['resume']) && !empty($modSettings['search_custom_index_resume']))
  470. {
  471. $context['index_settings'] = unserialize($modSettings['search_custom_index_resume']);
  472. $context['start'] = (int) $context['index_settings']['resume_at'];
  473. unset($context['index_settings']['resume_at']);
  474. $context['step'] = 1;
  475. }
  476. else
  477. {
  478. $context['index_settings'] = array(
  479. 'bytes_per_word' => isset($_REQUEST['bytes_per_word']) && isset($index_properties[$_REQUEST['bytes_per_word']]) ? (int) $_REQUEST['bytes_per_word'] : 2,
  480. );
  481. $context['start'] = isset($_REQUEST['start']) ? (int) $_REQUEST['start'] : 0;
  482. $context['step'] = isset($_REQUEST['step']) ? (int) $_REQUEST['step'] : 0;
  483. }
  484. if ($context['step'] !== 0)
  485. checkSession('request');
  486. // Step 0: let the user determine how they like their index.
  487. if ($context['step'] === 0)
  488. {
  489. $context['sub_template'] = 'create_index';
  490. }
  491. // Step 1: insert all the words.
  492. if ($context['step'] === 1)
  493. {
  494. $context['sub_template'] = 'create_index_progress';
  495. if ($context['start'] === 0)
  496. {
  497. db_extend();
  498. $tables = $smcFunc['db_list_tables'](false, $db_prefix . 'log_search_words');
  499. if (!empty($tables))
  500. {
  501. $smcFunc['db_search_query']('drop_words_table', '
  502. DROP TABLE {db_prefix}log_search_words',
  503. array(
  504. )
  505. );
  506. }
  507. $smcFunc['db_create_word_search']($index_properties[$context['index_settings']['bytes_per_word']]['column_definition']);
  508. // Temporarily switch back to not using a search index.
  509. if (!empty($modSettings['search_index']) && $modSettings['search_index'] == 'custom')
  510. updateSettings(array('search_index' => ''));
  511. // Don't let simultanious processes be updating the search index.
  512. if (!empty($modSettings['search_custom_index_config']))
  513. updateSettings(array('search_custom_index_config' => ''));
  514. }
  515. $num_messages = array(
  516. 'done' => 0,
  517. 'todo' => 0,
  518. );
  519. $request = $smcFunc['db_query']('', '
  520. SELECT id_msg >= {int:starting_id} AS todo, COUNT(*) AS num_messages
  521. FROM {db_prefix}messages
  522. GROUP BY todo',
  523. array(
  524. 'starting_id' => $context['start'],
  525. )
  526. );
  527. while ($row = $smcFunc['db_fetch_assoc']($request))
  528. $num_messages[empty($row['todo']) ? 'done' : 'todo'] = $row['num_messages'];
  529. if (empty($num_messages['todo']))
  530. {
  531. $context['step'] = 2;
  532. $context['percentage'] = 80;
  533. $context['start'] = 0;
  534. }
  535. else
  536. {
  537. // Number of seconds before the next step.
  538. $stop = time() + 3;
  539. while (time() < $stop)
  540. {
  541. $inserts = array();
  542. $request = $smcFunc['db_query']('', '
  543. SELECT id_msg, body
  544. FROM {db_prefix}messages
  545. WHERE id_msg BETWEEN {int:starting_id} AND {int:ending_id}
  546. LIMIT {int:limit}',
  547. array(
  548. 'starting_id' => $context['start'],
  549. 'ending_id' => $context['start'] + $messages_per_batch - 1,
  550. 'limit' => $messages_per_batch,
  551. )
  552. );
  553. $forced_break = false;
  554. $number_processed = 0;
  555. while ($row = $smcFunc['db_fetch_assoc']($request))
  556. {
  557. // In theory it's possible for one of these to take friggin ages so add more timeout protection.
  558. if ($stop < time())
  559. {
  560. $forced_break = true;
  561. break;
  562. }
  563. $number_processed++;
  564. foreach (text2words($row['body'], $context['index_settings']['bytes_per_word'], true) as $id_word)
  565. {
  566. $inserts[] = array($id_word, $row['id_msg']);
  567. }
  568. }
  569. $num_messages['done'] += $number_processed;
  570. $num_messages['todo'] -= $number_processed;
  571. $smcFunc['db_free_result']($request);
  572. $context['start'] += $forced_break ? $number_processed : $messages_per_batch;
  573. if (!empty($inserts))
  574. $smcFunc['db_insert']('ignore',
  575. '{db_prefix}log_search_words',
  576. array('id_word' => 'int', 'id_msg' => 'int'),
  577. $inserts,
  578. array('id_word', 'id_msg')
  579. );
  580. if ($num_messages['todo'] === 0)
  581. {
  582. $context['step'] = 2;
  583. $context['start'] = 0;
  584. break;
  585. }
  586. else
  587. updateSettings(array('search_custom_index_resume' => serialize(array_merge($context['index_settings'], array('resume_at' => $context['start'])))));
  588. }
  589. // Since there are still two steps to go, 90% is the maximum here.
  590. $context['percentage'] = round($num_messages['done'] / ($num_messages['done'] + $num_messages['todo']), 3) * 80;
  591. }
  592. }
  593. // Step 2: removing the words that occur too often and are of no use.
  594. elseif ($context['step'] === 2)
  595. {
  596. if ($context['index_settings']['bytes_per_word'] < 4)
  597. $context['step'] = 3;
  598. else
  599. {
  600. $stop_words = $context['start'] === 0 || empty($modSettings['search_stopwords']) ? array() : explode(',', $modSettings['search_stopwords']);
  601. $stop = time() + 3;
  602. $context['sub_template'] = 'create_index_progress';
  603. $max_messages = ceil(60 * $modSettings['totalMessages'] / 100);
  604. while (time() < $stop)
  605. {
  606. $request = $smcFunc['db_query']('', '
  607. SELECT id_word, COUNT(id_word) AS num_words
  608. FROM {db_prefix}log_search_words
  609. WHERE id_word BETWEEN {int:starting_id} AND {int:ending_id}
  610. GROUP BY id_word
  611. HAVING COUNT(id_word) > {int:minimum_messages}',
  612. array(
  613. 'starting_id' => $context['start'],
  614. 'ending_id' => $context['start'] + $index_properties[$context['index_settings']['bytes_per_word']]['step_size'] - 1,
  615. 'minimum_messages' => $max_messages,
  616. )
  617. );
  618. while ($row = $smcFunc['db_fetch_assoc']($request))
  619. $stop_words[] = $row['id_word'];
  620. $smcFunc['db_free_result']($request);
  621. updateSettings(array('search_stopwords' => implode(',', $stop_words)));
  622. if (!empty($stop_words))
  623. $smcFunc['db_query']('', '
  624. DELETE FROM {db_prefix}log_search_words
  625. WHERE id_word in ({array_int:stop_words})',
  626. array(
  627. 'stop_words' => $stop_words,
  628. )
  629. );
  630. $context['start'] += $index_properties[$context['index_settings']['bytes_per_word']]['step_size'];
  631. if ($context['start'] > $index_properties[$context['index_settings']['bytes_per_word']]['max_size'])
  632. {
  633. $context['step'] = 3;
  634. break;
  635. }
  636. }
  637. $context['percentage'] = 80 + round($context['start'] / $index_properties[$context['index_settings']['bytes_per_word']]['max_size'], 3) * 20;
  638. }
  639. }
  640. // Step 3: remove words not distinctive enough.
  641. if ($context['step'] === 3)
  642. {
  643. $context['sub_template'] = 'create_index_done';
  644. updateSettings(array('search_index' => 'custom', 'search_custom_index_config' => serialize($context['index_settings'])));
  645. $smcFunc['db_query']('', '
  646. DELETE FROM {db_prefix}settings
  647. WHERE variable = {string:search_custom_index_resume}',
  648. array(
  649. 'search_custom_index_resume' => 'search_custom_index_resume',
  650. )
  651. );
  652. }
  653. }
  654. /**
  655. * Get the installed Search API implementations.
  656. * This function checks for patterns in comments on top of the Search-API files!
  657. * In addition to filenames pattern.
  658. * It loads the search API classes if identified.
  659. * This function is used by EditSearchMethod to list all installed API implementations.
  660. */
  661. function loadSearchAPIs()
  662. {
  663. global $sourcedir, $txt;
  664. $apis = array();
  665. if ($dh = opendir($sourcedir))
  666. {
  667. while (($file = readdir($dh)) !== false)
  668. {
  669. if (is_file($sourcedir . '/' . $file) && preg_match('~SearchAPI-([A-Za-z\d_]+)\.php~', $file, $matches))
  670. {
  671. // Check this is definitely a valid API!
  672. $fp = fopen($sourcedir . '/' . $file, 'rb');
  673. $header = fread($fp, 4096);
  674. fclose($fp);
  675. if (strpos($header, '* SearchAPI-' . $matches[1] . '.php') !== false)
  676. {
  677. loadClassFile($file);
  678. $index_name = strtolower($matches[1]);
  679. $search_class_name = $index_name . '_search';
  680. $searchAPI = new $search_class_name();
  681. // No Support? NEXT!
  682. if (!$searchAPI->is_supported)
  683. continue;
  684. $apis[$index_name] = array(
  685. 'filename' => $file,
  686. 'setting_index' => $index_name,
  687. 'has_template' => in_array($index_name, array('custom', 'fulltext', 'standard')),
  688. 'label' => $index_name && isset($txt['search_index_' . $index_name]) ? $txt['search_index_' . $index_name] : '',
  689. 'desc' => $index_name && isset($txt['search_index_' . $index_name . '_desc']) ? $txt['search_index_' . $index_name . '_desc'] : '',
  690. );
  691. }
  692. }
  693. }
  694. }
  695. closedir($dh);
  696. return $apis;
  697. }
  698. ?>