| Index: trunk/extensions/BotQuery/query.php |
| — | — | @@ -75,35 +75,15 @@ |
| 76 | 76 | define( 'NS_ALL_NAMESPACES', -10123 ); |
| 77 | 77 | |
| 78 | 78 | |
| 79 | | -// Possible way to implement login functionality |
| 80 | | -//$action = $wgRequest->getVal('action', 'query'); |
| 81 | | -//switch( $action ) { |
| 82 | | -// case 'query': |
| 83 | | - $bqp = new BotQueryProcessor( $startTime ); |
| 84 | | - $bqp->execute(); |
| 85 | | - $bqp->output(); |
| 86 | | -// break; |
| 87 | | -// case 'login': |
| 88 | | -// $blp = new BotLoginProcessor(); |
| 89 | | -// $blp->execute(); |
| 90 | | -// break; |
| 91 | | -//} |
| | 79 | +$bqp = new BotQueryProcessor( $startTime ); |
| | 80 | +$bqp->execute(); |
| | 81 | +$bqp->output(); |
| 92 | 82 | |
| 93 | 83 | wfProfileOut( 'query.php' ); |
| 94 | 84 | if ( function_exists( 'wfLogProfilingData' ) ) { |
| 95 | 85 | wfLogProfilingData(); |
| 96 | 86 | } |
| 97 | 87 | |
| 98 | | -//class BotLoginProcessor { |
| 99 | | -// function BotLoginProcessor() |
| 100 | | -// {} |
| 101 | | -// |
| 102 | | -// function execute() |
| 103 | | -// { |
| 104 | | -// wfSetupSession(); |
| 105 | | -// } |
| 106 | | -//} |
| 107 | | - |
| 108 | 88 | class BotQueryProcessor { |
| 109 | 89 | var $classname = 'BotQueryProcessor'; |
| 110 | 90 | |
| — | — | @@ -502,15 +482,18 @@ |
| 503 | 483 | "ucrbtoken - If logged in as an admin, a rollback tokens for top revisions will be included in the output.", |
| 504 | 484 | "Example: query.php?what=usercontribs&titles=User:YurikBot&uclimit=20&uccomments", |
| 505 | 485 | )), |
| 506 | | - 'contribcounter' => array( |
| 507 | | - GN_FUNC => 'genContributionsCounter', |
| 508 | | - GN_ISMETA => false, |
| 509 | | - GN_PARAMS => array(), |
| 510 | | - GN_DFLT => array(), |
| 511 | | - GN_DESC => array( |
| 512 | | - "User contributions counter", |
| 513 | | - "Example: query.php?what=contribcounter&titles=User:Yurik", |
| 514 | | - )), |
| | 486 | + |
| | 487 | + // Extremelly slow query!!! |
| | 488 | + |
| | 489 | +// 'contribcounter' => array( |
| | 490 | +// GN_FUNC => 'genContributionsCounter', |
| | 491 | +// GN_ISMETA => false, |
| | 492 | +// GN_PARAMS => array(), |
| | 493 | +// GN_DFLT => array(), |
| | 494 | +// GN_DESC => array( |
| | 495 | +// "User contributions counter", |
| | 496 | +// "Example: query.php?what=contribcounter&titles=User:Yurik", |
| | 497 | +// )), |
| 515 | 498 | 'imageinfo' => array( |
| 516 | 499 | GN_FUNC => 'genImageInfo', |
| 517 | 500 | GN_ISMETA => false, |
| — | — | @@ -544,6 +527,9 @@ |
| 545 | 528 | { |
| 546 | 529 | global $wgRequest, $wgUser; |
| 547 | 530 | |
| | 531 | + // Initialize Error handler |
| | 532 | + set_exception_handler( array($this, 'ExceptionHandler') ); |
| | 533 | + |
| 548 | 534 | $this->startTime = $startTime; |
| 549 | 535 | $this->totalDbTime = 0; |
| 550 | 536 | |
| — | — | @@ -581,6 +567,40 @@ |
| 582 | 568 | } |
| 583 | 569 | |
| 584 | 570 | /** |
| | 571 | + * Exception handler which simulates the appropriate catch() handling: |
| | 572 | + * |
| | 573 | + * try { |
| | 574 | + * ... |
| | 575 | + * } catch ( MWException $e ) { |
| | 576 | + * dieUsage() |
| | 577 | + * } catch ( Exception $e ) { |
| | 578 | + * echo $e->__toString(); |
| | 579 | + * } |
| | 580 | + */ |
| | 581 | + function ExceptionHandler( $e ) { |
| | 582 | + global $wgFullyInitialised; |
| | 583 | + if ( is_a( $e, 'MWException' ) ) { |
| | 584 | + try { |
| | 585 | + $this->dieUsage( "Exception Caught: {$e->getMessage()}\n\n{$e->getTraceAsString()}\n\n", 'internal_error' ); |
| | 586 | + } catch (Exception $e2) { |
| | 587 | + echo $e->__toString(); |
| | 588 | + } |
| | 589 | + } else { |
| | 590 | + echo $e->__toString(); |
| | 591 | + } |
| | 592 | + |
| | 593 | + // Final cleanup, similar to wfErrorExit() |
| | 594 | + if ( $wgFullyInitialised ) { |
| | 595 | + try { |
| | 596 | + wfLogProfilingData(); // uses $wgRequest, hence the $wgFullyInitialised condition |
| | 597 | + } catch ( Exception $e ) {} |
| | 598 | + } |
| | 599 | + |
| | 600 | + // Exit value should be nonzero for the benefit of shell jobs |
| | 601 | + exit( 1 ); |
| | 602 | + } |
| | 603 | + |
| | 604 | + /** |
| 585 | 605 | * The core function - executes meta generators, populates basic page info, and then fills in the required additional data for all pages |
| 586 | 606 | */ |
| 587 | 607 | function execute() |
| — | — | @@ -1141,16 +1161,82 @@ |
| 1142 | 1162 | // |
| 1143 | 1163 | // ************************************* PAGE INFO GENERATORS ************************************* |
| 1144 | 1164 | // |
| 1145 | | - |
| | 1165 | + |
| 1146 | 1166 | /** |
| 1147 | | - * Populate redirect data. Redirects may be one of the following: |
| 1148 | | - * Redir to nonexisting, Existing page, or Existing redirect. |
| 1149 | | - * Existing redirect may point to yet another nonexisting or existing page( which in turn may also be a redirect) |
| 1150 | | - */ |
| | 1167 | + * Simpler replacement for double-redirect resolving genRedirectInfo2() |
| | 1168 | + */ |
| 1151 | 1169 | function genRedirectInfo(&$prop, &$genInfo) |
| 1152 | 1170 | { |
| 1153 | 1171 | if( empty( $this->redirectPageIds )) return; |
| 1154 | 1172 | $this->startProfiling(); |
| | 1173 | + |
| | 1174 | + $this->startDbProfiling(); |
| | 1175 | + $res = $this->db->select( |
| | 1176 | + array('page', 'pagelinks'), |
| | 1177 | + array('pl_from', 'pl_namespace', 'pl_title', 'page_id', 'page_is_redirect'), |
| | 1178 | + array('pl_from' => $this->redirectPageIds, |
| | 1179 | + 'pl_namespace = page_namespace', |
| | 1180 | + 'pl_title = page_title'), |
| | 1181 | + __CLASS__ . '::' . __FUNCTION__ ); |
| | 1182 | + $this->endDbProfiling( $prop ); |
| | 1183 | + |
| | 1184 | + $multiLinkRedirPages = array(); |
| | 1185 | + |
| | 1186 | + while ( $row = $this->db->fetchObject( $res ) ) { |
| | 1187 | + $pageId = intval($row->pl_from); |
| | 1188 | + $data = & $this->data['pages'][$pageId]['redirect']; |
| | 1189 | + if ( !isset($data['to']) ) { |
| | 1190 | + $data['to'] = $this->getLinkInfo( $row->pl_namespace, $row->pl_title, $row->page_id, $row->page_is_redirect ); |
| | 1191 | + } else { |
| | 1192 | + // More than one link exists from redirect page |
| | 1193 | + $multiLinkRedirPages[$pageId] = ''; |
| | 1194 | + } |
| | 1195 | + } |
| | 1196 | + $this->db->freeResult( $res ); |
| | 1197 | + |
| | 1198 | + if (!empty($multiLinkRedirPages)) { |
| | 1199 | + // We found some bad redirect pages. Get the content and solve. |
| | 1200 | + $multiLinkRedirPages = array_keys($multiLinkRedirPages); |
| | 1201 | + $ids = array(); |
| | 1202 | + foreach( $multiLinkRedirPages as $pageId ) { |
| | 1203 | + $ids[] = "(rev_page=$pageId AND rev_id={$this->data['pages'][$pageId]['revid']})"; |
| | 1204 | + } |
| | 1205 | + |
| | 1206 | + $this->startDbProfiling(); |
| | 1207 | + $res = $this->db->select( |
| | 1208 | + array('page', 'revision', 'text'), |
| | 1209 | + array('page_id', 'page_is_redirect', 'old_id', 'old_text', 'old_flags'), |
| | 1210 | + array('page_id' => $multiLinkRedirPages, 'page_latest=rev_id', 'rev_text_id=old_id' ), |
| | 1211 | + $this->classname . '::genPageContent' |
| | 1212 | + ); |
| | 1213 | + while ( $row = $this->db->fetchObject( $res ) ) { |
| | 1214 | + $title = Title :: newFromRedirect(Revision::getRevisionText( $row )); |
| | 1215 | + if ($title) { |
| | 1216 | + $article = new Article($title); |
| | 1217 | + $pageId = $article->getTitle()->getArticleId(); |
| | 1218 | + $isRedirect = $pageId > 0 ? !$article->checkTouched() : false; |
| | 1219 | + $link = $this->getTitleInfo( $title, $pageId, $isRedirect ); |
| | 1220 | + $this->data['pages'][intval($row->page_id)]['redirect']['to'] = $link; |
| | 1221 | + } |
| | 1222 | + } |
| | 1223 | + $this->db->freeResult( $res ); |
| | 1224 | + } |
| | 1225 | + |
| | 1226 | + $this->endProfiling( $prop ); |
| | 1227 | + } |
| | 1228 | + |
| | 1229 | + /** |
| | 1230 | + * |
| | 1231 | + * This method cannot be used until http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 is fixed |
| | 1232 | + * |
| | 1233 | + * Populate redirect data. Redirects may be one of the following: |
| | 1234 | + * Redir to nonexisting, Existing page, or Existing redirect. |
| | 1235 | + * Existing redirect may point to yet another nonexisting or existing page( which in turn may also be a redirect) |
| | 1236 | + */ |
| | 1237 | + function genRedirectInfo2(&$prop, &$genInfo) |
| | 1238 | + { |
| | 1239 | + if( empty( $this->redirectPageIds )) return; |
| | 1240 | + $this->startProfiling(); |
| 1155 | 1241 | extract( $this->db->tableNames( 'page', 'pagelinks' ) ); |
| 1156 | 1242 | |
| 1157 | 1243 | // |
| — | — | @@ -1815,45 +1901,45 @@ |
| 1816 | 1902 | } |
| 1817 | 1903 | |
| 1818 | 1904 | |
| 1819 | | - /** |
| 1820 | | - * Add counts of user contributions to the user pages |
| 1821 | | - */ |
| 1822 | | - function genContributionsCounter(&$prop, &$genInfo) |
| 1823 | | - { |
| 1824 | | - $this->startProfiling(); |
| 1825 | | - $users = array (); // Users to query |
| 1826 | | - $userPageIds = array (); // Map of user name to the page ID |
| 1827 | | - |
| 1828 | | - // For all valid pages in User namespace query history. Note that the page might not exist. |
| 1829 | | - foreach ($this->data['pages'] as $pageId => & $page) { |
| 1830 | | - if (array_key_exists('_obj', $page)) { |
| 1831 | | - $title = & $page['_obj']; |
| 1832 | | - if ($title->getNamespace() == NS_USER && !$title->isExternal()) { |
| 1833 | | - $users[] = $title->getText(); |
| 1834 | | - $userPageIds[$title->getText()] = $pageId; |
| 1835 | | - } |
| 1836 | | - } |
| 1837 | | - } |
| 1838 | | - |
| 1839 | | - $this->validateLimit( 'cc_querytoobig', count($users), 10, 50 ); |
| 1840 | | - $this->startDbProfiling(); |
| 1841 | | - $res = $this->db->select('revision', array ( |
| 1842 | | - 'rev_user_text', |
| 1843 | | - 'count(*) cnt', |
| 1844 | | - 'count(DISTINCT rev_page) distcnt' |
| 1845 | | - ), array ( |
| 1846 | | - 'rev_user_text' => $users |
| 1847 | | - ), $this->classname . '::genContributionsCounter', array ( |
| 1848 | | - 'GROUP BY' => 'rev_user_text' |
| 1849 | | - )); |
| 1850 | | - $this->endDbProfiling($prop); |
| 1851 | | - while ($row = $this->db->fetchObject($res)) { |
| 1852 | | - $pageId = $userPageIds[$row->rev_user_text]; |
| 1853 | | - $this->addPageSubElement($pageId, $prop, 'count', $row->cnt, false); |
| 1854 | | - $this->addPageSubElement($pageId, $prop, 'distcount', $row->distcnt, false); |
| 1855 | | - } |
| 1856 | | - $this->endProfiling($prop); |
| 1857 | | - } |
| | 1905 | +// /** |
| | 1906 | +// * Add counts of user contributions to the user pages |
| | 1907 | +// */ |
| | 1908 | +// function genContributionsCounter(&$prop, &$genInfo) |
| | 1909 | +// { |
| | 1910 | +// $this->startProfiling(); |
| | 1911 | +// $users = array (); // Users to query |
| | 1912 | +// $userPageIds = array (); // Map of user name to the page ID |
| | 1913 | +// |
| | 1914 | +// // For all valid pages in User namespace query history. Note that the page might not exist. |
| | 1915 | +// foreach ($this->data['pages'] as $pageId => & $page) { |
| | 1916 | +// if (array_key_exists('_obj', $page)) { |
| | 1917 | +// $title = & $page['_obj']; |
| | 1918 | +// if ($title->getNamespace() == NS_USER && !$title->isExternal()) { |
| | 1919 | +// $users[] = $title->getText(); |
| | 1920 | +// $userPageIds[$title->getText()] = $pageId; |
| | 1921 | +// } |
| | 1922 | +// } |
| | 1923 | +// } |
| | 1924 | +// |
| | 1925 | +// $this->validateLimit( 'cc_querytoobig', count($users), 10, 50 ); |
| | 1926 | +// $this->startDbProfiling(); |
| | 1927 | +// $res = $this->db->select('revision', array ( |
| | 1928 | +// 'rev_user_text', |
| | 1929 | +// 'count(*) cnt', |
| | 1930 | +// 'count(DISTINCT rev_page) distcnt' |
| | 1931 | +// ), array ( |
| | 1932 | +// 'rev_user_text' => $users |
| | 1933 | +// ), $this->classname . '::genContributionsCounter', array ( |
| | 1934 | +// 'GROUP BY' => 'rev_user_text' |
| | 1935 | +// )); |
| | 1936 | +// $this->endDbProfiling($prop); |
| | 1937 | +// while ($row = $this->db->fetchObject($res)) { |
| | 1938 | +// $pageId = $userPageIds[$row->rev_user_text]; |
| | 1939 | +// $this->addPageSubElement($pageId, $prop, 'count', $row->cnt, false); |
| | 1940 | +// $this->addPageSubElement($pageId, $prop, 'distcount', $row->distcnt, false); |
| | 1941 | +// } |
| | 1942 | +// $this->endProfiling($prop); |
| | 1943 | +// } |
| 1858 | 1944 | |
| 1859 | 1945 | /** |
| 1860 | 1946 | * Add the raw content of the pages |