Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
0.00% |
0 / 323 |
|
0.00% |
0 / 9 |
CRAP | |
0.00% |
0 / 1 |
| RebuildRecentchanges | |
0.00% |
0 / 323 |
|
0.00% |
0 / 9 |
2256 | |
0.00% |
0 / 1 |
| __construct | |
0.00% |
0 / 15 |
|
0.00% |
0 / 1 |
2 | |||
| execute | |
0.00% |
0 / 11 |
|
0.00% |
0 / 1 |
56 | |||
| rebuildRecentChangesTablePass1 | |
0.00% |
0 / 86 |
|
0.00% |
0 / 1 |
56 | |||
| rebuildRecentChangesTablePass2 | |
0.00% |
0 / 55 |
|
0.00% |
0 / 1 |
72 | |||
| rebuildRecentChangesTablePass3 | |
0.00% |
0 / 73 |
|
0.00% |
0 / 1 |
30 | |||
| findRcIdsWithGroups | |
0.00% |
0 / 16 |
|
0.00% |
0 / 1 |
6 | |||
| rebuildRecentChangesTablePass4 | |
0.00% |
0 / 34 |
|
0.00% |
0 / 1 |
156 | |||
| rebuildRecentChangesTablePass5 | |
0.00% |
0 / 28 |
|
0.00% |
0 / 1 |
12 | |||
| purgeFeeds | |
0.00% |
0 / 5 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | /** |
| 3 | * Rebuild recent changes from scratch. This takes several hours, |
| 4 | * depending on the database size and server configuration. |
| 5 | * |
| 6 | * @license GPL-2.0-or-later |
| 7 | * @file |
| 8 | * @ingroup Maintenance |
| 9 | * @todo Document |
| 10 | */ |
| 11 | |
| 12 | // @codeCoverageIgnoreStart |
| 13 | require_once __DIR__ . '/Maintenance.php'; |
| 14 | // @codeCoverageIgnoreEnd |
| 15 | |
| 16 | use MediaWiki\Logging\LogPage; |
| 17 | use MediaWiki\Maintenance\Maintenance; |
| 18 | use MediaWiki\RecentChanges\RecentChange; |
| 19 | use Wikimedia\Rdbms\IDatabase; |
| 20 | use Wikimedia\Rdbms\IReadableDatabase; |
| 21 | use Wikimedia\Rdbms\SelectQueryBuilder; |
| 22 | use Wikimedia\Timestamp\TimestampFormat as TS; |
| 23 | |
| 24 | /** |
| 25 | * Maintenance script that rebuilds recent changes from scratch. |
| 26 | * |
| 27 | * @ingroup Maintenance |
| 28 | */ |
| 29 | class RebuildRecentchanges extends Maintenance { |
| 30 | /** @var int UNIX timestamp */ |
| 31 | private $cutoffFrom; |
| 32 | /** @var int UNIX timestamp */ |
| 33 | private $cutoffTo; |
| 34 | |
| 35 | public function __construct() { |
| 36 | parent::__construct(); |
| 37 | $this->addDescription( 'Rebuild recent changes' ); |
| 38 | |
| 39 | $this->addOption( |
| 40 | 'from', |
| 41 | "Only rebuild rows in requested time range (in YYYYMMDDHHMMSS format)", |
| 42 | false, |
| 43 | true |
| 44 | ); |
| 45 | $this->addOption( |
| 46 | 'to', |
| 47 | "Only rebuild rows in requested time range (in YYYYMMDDHHMMSS format)", |
| 48 | false, |
| 49 | true |
| 50 | ); |
| 51 | $this->setBatchSize( 200 ); |
| 52 | } |
| 53 | |
| 54 | public function execute() { |
| 55 | if ( |
| 56 | ( $this->hasOption( 'from' ) && !$this->hasOption( 'to' ) ) || |
| 57 | ( !$this->hasOption( 'from' ) && $this->hasOption( 'to' ) ) |
| 58 | ) { |
| 59 | $this->fatalError( "Both 'from' and 'to' must be given, or neither" ); |
| 60 | } |
| 61 | |
| 62 | $this->rebuildRecentChangesTablePass1(); |
| 63 | $this->rebuildRecentChangesTablePass2(); |
| 64 | $this->rebuildRecentChangesTablePass3(); |
| 65 | $this->rebuildRecentChangesTablePass4(); |
| 66 | $this->rebuildRecentChangesTablePass5(); |
| 67 | if ( !( $this->hasOption( 'from' ) && $this->hasOption( 'to' ) ) ) { |
| 68 | $this->purgeFeeds(); |
| 69 | } |
| 70 | $this->output( "Done.\n" ); |
| 71 | } |
| 72 | |
| 73 | /** |
| 74 | * Rebuild pass 1: Insert `recentchanges` entries for page revisions. |
| 75 | */ |
| 76 | private function rebuildRecentChangesTablePass1() { |
| 77 | $dbw = $this->getPrimaryDB(); |
| 78 | $commentStore = $this->getServiceContainer()->getCommentStore(); |
| 79 | |
| 80 | if ( $this->hasOption( 'from' ) && $this->hasOption( 'to' ) ) { |
| 81 | $this->cutoffFrom = (int)wfTimestamp( TS::UNIX, $this->getOption( 'from' ) ); |
| 82 | $this->cutoffTo = (int)wfTimestamp( TS::UNIX, $this->getOption( 'to' ) ); |
| 83 | |
| 84 | $sec = $this->cutoffTo - $this->cutoffFrom; |
| 85 | $days = $sec / 24 / 3600; |
| 86 | $this->output( "Rebuilding range of $sec seconds ($days days)\n" ); |
| 87 | } else { |
| 88 | global $wgRCMaxAge; |
| 89 | |
| 90 | $days = $wgRCMaxAge / 24 / 3600; |
| 91 | $this->output( "Rebuilding \$wgRCMaxAge=$wgRCMaxAge seconds ($days days)\n" ); |
| 92 | |
| 93 | $this->cutoffFrom = time() - $wgRCMaxAge; |
| 94 | $this->cutoffTo = time(); |
| 95 | } |
| 96 | |
| 97 | $this->output( "Clearing recentchanges table for time range...\n" ); |
| 98 | $rcids = $dbw->newSelectQueryBuilder() |
| 99 | ->select( 'rc_id' ) |
| 100 | ->from( 'recentchanges' ) |
| 101 | ->where( $dbw->expr( 'rc_timestamp', '>', $dbw->timestamp( $this->cutoffFrom ) ) ) |
| 102 | ->andWhere( $dbw->expr( 'rc_timestamp', '<', $dbw->timestamp( $this->cutoffTo ) ) ) |
| 103 | ->caller( __METHOD__ )->fetchFieldValues(); |
| 104 | foreach ( array_chunk( $rcids, $this->getBatchSize() ) as $rcidBatch ) { |
| 105 | $dbw->newDeleteQueryBuilder() |
| 106 | ->deleteFrom( 'recentchanges' ) |
| 107 | ->where( [ 'rc_id' => $rcidBatch ] ) |
| 108 | ->caller( __METHOD__ )->execute(); |
| 109 | $this->waitForReplication(); |
| 110 | } |
| 111 | |
| 112 | $this->output( "Loading from page and revision tables...\n" ); |
| 113 | |
| 114 | $res = $dbw->newSelectQueryBuilder() |
| 115 | ->select( |
| 116 | [ |
| 117 | 'rev_timestamp', |
| 118 | 'rev_minor_edit', |
| 119 | 'rev_id', |
| 120 | 'rev_deleted', |
| 121 | 'page_namespace', |
| 122 | 'page_title', |
| 123 | 'page_is_new', |
| 124 | 'page_id', |
| 125 | 'rev_comment_text' => 'comment_rev_comment.comment_text', |
| 126 | 'rev_comment_data' => 'comment_rev_comment.comment_data', |
| 127 | 'rev_comment_cid' => 'comment_rev_comment.comment_id', |
| 128 | 'rev_user' => 'actor_rev_user.actor_user', |
| 129 | 'rev_user_text' => 'actor_rev_user.actor_name', |
| 130 | 'rev_actor' => 'rev_actor', |
| 131 | ] |
| 132 | ) |
| 133 | ->from( 'revision' ) |
| 134 | ->join( 'page', null, 'rev_page=page_id' ) |
| 135 | ->join( 'comment', 'comment_rev_comment', 'comment_rev_comment.comment_id = rev_comment_id' ) |
| 136 | ->join( 'actor', 'actor_rev_user', 'actor_rev_user.actor_id = rev_actor' ) |
| 137 | ->where( |
| 138 | [ |
| 139 | $dbw->expr( 'rev_timestamp', '>', $dbw->timestamp( $this->cutoffFrom ) ), |
| 140 | $dbw->expr( 'rev_timestamp', '<', $dbw->timestamp( $this->cutoffTo ) ) |
| 141 | ] |
| 142 | ) |
| 143 | ->orderBy( 'rev_timestamp', SelectQueryBuilder::SORT_DESC ) |
| 144 | ->caller( __METHOD__ )->fetchResultSet(); |
| 145 | |
| 146 | $this->output( "Inserting from page and revision tables...\n" ); |
| 147 | $inserted = 0; |
| 148 | foreach ( $res as $row ) { |
| 149 | $comment = $commentStore->getComment( 'rev_comment', $row ); |
| 150 | $dbw->newInsertQueryBuilder() |
| 151 | ->insertInto( 'recentchanges' ) |
| 152 | ->row( [ |
| 153 | 'rc_timestamp' => $row->rev_timestamp, |
| 154 | 'rc_actor' => $row->rev_actor, |
| 155 | 'rc_namespace' => $row->page_namespace, |
| 156 | 'rc_title' => $row->page_title, |
| 157 | 'rc_minor' => $row->rev_minor_edit, |
| 158 | 'rc_bot' => 0, |
| 159 | 'rc_cur_id' => $row->page_id, |
| 160 | 'rc_this_oldid' => $row->rev_id, |
| 161 | 'rc_last_oldid' => 0, // is this ok? |
| 162 | 'rc_source' => $row->page_is_new ? RecentChange::SRC_NEW : RecentChange::SRC_EDIT, |
| 163 | 'rc_deleted' => $row->rev_deleted |
| 164 | ] + $commentStore->insert( $dbw, 'rc_comment', $comment ) ) |
| 165 | ->caller( __METHOD__ )->execute(); |
| 166 | |
| 167 | $rcid = $dbw->insertId(); |
| 168 | $dbw->newUpdateQueryBuilder() |
| 169 | ->update( 'change_tag' ) |
| 170 | ->set( [ 'ct_rc_id' => $rcid ] ) |
| 171 | ->where( [ 'ct_rev_id' => $row->rev_id ] ) |
| 172 | ->caller( __METHOD__ )->execute(); |
| 173 | |
| 174 | if ( ( ++$inserted % $this->getBatchSize() ) == 0 ) { |
| 175 | $this->waitForReplication(); |
| 176 | } |
| 177 | } |
| 178 | } |
| 179 | |
| 180 | /** |
| 181 | * Rebuild pass 2: Enhance entries for page revisions with references to the previous revision |
| 182 | * (rc_last_oldid etc.) and size differences (rc_old_len, rc_new_len). |
| 183 | */ |
| 184 | private function rebuildRecentChangesTablePass2() { |
| 185 | $dbw = $this->getPrimaryDB(); |
| 186 | |
| 187 | $this->output( "Updating links and size differences...\n" ); |
| 188 | |
| 189 | # Fill in the rc_last_oldid field, which points to the previous edit |
| 190 | $res = $dbw->newSelectQueryBuilder() |
| 191 | ->select( [ 'rc_cur_id', 'rc_this_oldid', 'rc_timestamp' ] ) |
| 192 | ->from( 'recentchanges' ) |
| 193 | ->where( $dbw->expr( 'rc_timestamp', '>', $dbw->timestamp( $this->cutoffFrom ) ) ) |
| 194 | ->andWhere( $dbw->expr( 'rc_timestamp', '<', $dbw->timestamp( $this->cutoffTo ) ) ) |
| 195 | ->orderBy( [ 'rc_cur_id', 'rc_timestamp' ] ) |
| 196 | ->caller( __METHOD__ )->fetchResultSet(); |
| 197 | |
| 198 | $lastCurId = 0; |
| 199 | $lastOldId = 0; |
| 200 | $lastSize = null; |
| 201 | $updated = 0; |
| 202 | foreach ( $res as $row ) { |
| 203 | $new = 0; |
| 204 | |
| 205 | if ( $row->rc_cur_id != $lastCurId ) { |
| 206 | # Switch! Look up the previous last edit, if any |
| 207 | $lastCurId = intval( $row->rc_cur_id ); |
| 208 | $emit = $row->rc_timestamp; |
| 209 | |
| 210 | $revRow = $dbw->newSelectQueryBuilder() |
| 211 | ->select( [ 'rev_id', 'rev_len' ] ) |
| 212 | ->from( 'revision' ) |
| 213 | ->where( [ 'rev_page' => $lastCurId, $dbw->expr( 'rev_timestamp', '<', $emit ) ] ) |
| 214 | ->orderBy( 'rev_timestamp DESC' ) |
| 215 | ->caller( __METHOD__ )->fetchRow(); |
| 216 | if ( $revRow ) { |
| 217 | $lastOldId = intval( $revRow->rev_id ); |
| 218 | # Grab the last text size if available |
| 219 | $lastSize = $revRow->rev_len !== null ? intval( $revRow->rev_len ) : null; |
| 220 | } else { |
| 221 | # No previous edit |
| 222 | $lastOldId = 0; |
| 223 | $lastSize = 0; |
| 224 | $new = 1; // probably true |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | if ( $lastCurId == 0 ) { |
| 229 | $this->output( "Uhhh, something wrong? No curid\n" ); |
| 230 | } else { |
| 231 | # Grab the entry's text size |
| 232 | $size = (int)$dbw->newSelectQueryBuilder() |
| 233 | ->select( 'rev_len' ) |
| 234 | ->from( 'revision' ) |
| 235 | ->where( [ 'rev_id' => $row->rc_this_oldid ] ) |
| 236 | ->caller( __METHOD__ )->fetchField(); |
| 237 | |
| 238 | $dbw->newUpdateQueryBuilder() |
| 239 | ->update( 'recentchanges' ) |
| 240 | ->set( [ |
| 241 | 'rc_last_oldid' => $lastOldId, |
| 242 | 'rc_source' => $new === 1 ? RecentChange::SRC_NEW : RecentChange::SRC_EDIT, |
| 243 | 'rc_old_len' => $lastSize, |
| 244 | 'rc_new_len' => $size, |
| 245 | ] ) |
| 246 | ->where( [ |
| 247 | 'rc_cur_id' => $lastCurId, |
| 248 | 'rc_this_oldid' => $row->rc_this_oldid, |
| 249 | // index usage |
| 250 | 'rc_timestamp' => $row->rc_timestamp, |
| 251 | ] ) |
| 252 | ->caller( __METHOD__ )->execute(); |
| 253 | |
| 254 | $lastOldId = intval( $row->rc_this_oldid ); |
| 255 | $lastSize = $size; |
| 256 | |
| 257 | if ( ( ++$updated % $this->getBatchSize() ) == 0 ) { |
| 258 | $this->waitForReplication(); |
| 259 | } |
| 260 | } |
| 261 | } |
| 262 | } |
| 263 | |
| 264 | /** |
| 265 | * Rebuild pass 3: Insert `recentchanges` entries for action logs. |
| 266 | */ |
| 267 | private function rebuildRecentChangesTablePass3() { |
| 268 | global $wgLogRestrictions, $wgFilterLogTypes; |
| 269 | |
| 270 | $dbw = $this->getDB( DB_PRIMARY ); |
| 271 | $commentStore = $this->getServiceContainer()->getCommentStore(); |
| 272 | $nonRCLogs = array_merge( |
| 273 | array_keys( $wgLogRestrictions ), |
| 274 | array_keys( $wgFilterLogTypes ), |
| 275 | [ 'create' ] |
| 276 | ); |
| 277 | |
| 278 | $this->output( "Loading from user and logging tables...\n" ); |
| 279 | |
| 280 | $res = $dbw->newSelectQueryBuilder() |
| 281 | ->select( |
| 282 | [ |
| 283 | 'log_timestamp', |
| 284 | 'log_actor', |
| 285 | 'log_namespace', |
| 286 | 'log_title', |
| 287 | 'log_page', |
| 288 | 'log_type', |
| 289 | 'log_action', |
| 290 | 'log_id', |
| 291 | 'log_params', |
| 292 | 'log_deleted', |
| 293 | 'log_comment_text' => 'comment_log_comment.comment_text', |
| 294 | 'log_comment_data' => 'comment_log_comment.comment_data', |
| 295 | 'log_comment_cid' => 'comment_log_comment.comment_id', |
| 296 | ] |
| 297 | ) |
| 298 | ->from( 'logging' ) |
| 299 | ->join( 'comment', 'comment_log_comment', 'comment_log_comment.comment_id = log_comment_id' ) |
| 300 | ->where( |
| 301 | [ |
| 302 | $dbw->expr( 'log_timestamp', '>', $dbw->timestamp( $this->cutoffFrom ) ), |
| 303 | $dbw->expr( 'log_timestamp', '<', $dbw->timestamp( $this->cutoffTo ) ), |
| 304 | // Some logs don't go in RC since they are private, or are included in the filterable log types. |
| 305 | 'log_type' => array_diff( LogPage::validTypes(), $nonRCLogs ), |
| 306 | ] |
| 307 | ) |
| 308 | ->orderBy( [ 'log_timestamp DESC', 'log_id DESC' ] ) |
| 309 | ->caller( __METHOD__ )->fetchResultSet(); |
| 310 | |
| 311 | $field = $dbw->fieldInfo( 'recentchanges', 'rc_cur_id' ); |
| 312 | |
| 313 | $inserted = 0; |
| 314 | foreach ( $res as $row ) { |
| 315 | $comment = $commentStore->getComment( 'log_comment', $row ); |
| 316 | $dbw->newInsertQueryBuilder() |
| 317 | ->insertInto( 'recentchanges' ) |
| 318 | ->row( [ |
| 319 | 'rc_timestamp' => $row->log_timestamp, |
| 320 | 'rc_actor' => $row->log_actor, |
| 321 | 'rc_namespace' => $row->log_namespace, |
| 322 | 'rc_title' => $row->log_title, |
| 323 | 'rc_minor' => 0, |
| 324 | 'rc_bot' => 0, |
| 325 | 'rc_patrolled' => $row->log_type == 'upload' ? 0 : 2, |
| 326 | 'rc_this_oldid' => 0, |
| 327 | 'rc_last_oldid' => 0, |
| 328 | 'rc_source' => RecentChange::SRC_LOG, |
| 329 | 'rc_cur_id' => $field->isNullable() |
| 330 | ? $row->log_page |
| 331 | : (int)$row->log_page, // NULL => 0, |
| 332 | 'rc_log_type' => $row->log_type, |
| 333 | 'rc_log_action' => $row->log_action, |
| 334 | 'rc_logid' => $row->log_id, |
| 335 | 'rc_params' => $row->log_params, |
| 336 | 'rc_deleted' => $row->log_deleted |
| 337 | ] + $commentStore->insert( $dbw, 'rc_comment', $comment ) ) |
| 338 | ->caller( __METHOD__ )->execute(); |
| 339 | |
| 340 | $rcid = $dbw->insertId(); |
| 341 | $dbw->newUpdateQueryBuilder() |
| 342 | ->update( 'change_tag' ) |
| 343 | ->set( [ 'ct_rc_id' => $rcid ] ) |
| 344 | ->where( [ 'ct_log_id' => $row->log_id ] ) |
| 345 | ->caller( __METHOD__ )->execute(); |
| 346 | |
| 347 | if ( ( ++$inserted % $this->getBatchSize() ) == 0 ) { |
| 348 | $this->waitForReplication(); |
| 349 | } |
| 350 | } |
| 351 | } |
| 352 | |
| 353 | /** |
| 354 | * Find rc_id values that have a user with one of the specified groups |
| 355 | * |
| 356 | * @param IReadableDatabase $db |
| 357 | * @param string[] $groups |
| 358 | * @param array $conds Extra query conditions |
| 359 | * @return int[] |
| 360 | */ |
| 361 | private function findRcIdsWithGroups( $db, $groups, $conds = [] ) { |
| 362 | if ( !count( $groups ) ) { |
| 363 | return []; |
| 364 | } |
| 365 | return $db->newSelectQueryBuilder() |
| 366 | ->select( 'rc_id' ) |
| 367 | ->distinct() |
| 368 | ->from( 'recentchanges' ) |
| 369 | ->join( 'actor', null, 'actor_id=rc_actor' ) |
| 370 | ->join( 'user_groups', null, 'ug_user=actor_user' ) |
| 371 | ->where( $conds ) |
| 372 | ->andWhere( [ |
| 373 | $db->expr( 'rc_timestamp', '>', $db->timestamp( $this->cutoffFrom ) ), |
| 374 | $db->expr( 'rc_timestamp', '<', $db->timestamp( $this->cutoffTo ) ), |
| 375 | 'ug_group' => $groups |
| 376 | ] ) |
| 377 | ->caller( __METHOD__ )->fetchFieldValues(); |
| 378 | } |
| 379 | |
| 380 | /** |
| 381 | * Rebuild pass 4: Mark bot and autopatrolled entries. |
| 382 | */ |
| 383 | private function rebuildRecentChangesTablePass4() { |
| 384 | global $wgUseRCPatrol, $wgUseNPPatrol, $wgUseFilePatrol, $wgMiserMode; |
| 385 | |
| 386 | $dbw = $this->getPrimaryDB(); |
| 387 | |
| 388 | # @FIXME: recognize other bot account groups (not the same as users with 'bot' rights) |
| 389 | # @NOTE: users with 'bot' rights choose when edits are bot edits or not. That information |
| 390 | # may be lost at this point (aside from joining on the patrol log table entries). |
| 391 | $botgroups = [ 'bot' ]; |
| 392 | $autopatrolgroups = ( $wgUseRCPatrol || $wgUseNPPatrol || $wgUseFilePatrol ) ? |
| 393 | $this->getServiceContainer()->getGroupPermissionsLookup() |
| 394 | ->getGroupsWithPermission( 'autopatrol' ) : []; |
| 395 | |
| 396 | # Flag our recent bot edits |
| 397 | // @phan-suppress-next-line PhanRedundantCondition |
| 398 | if ( $botgroups ) { |
| 399 | $this->output( "Flagging bot account edits...\n" ); |
| 400 | |
| 401 | # Fill in the rc_bot field |
| 402 | $rcids = $this->findRcIdsWithGroups( $dbw, $botgroups ); |
| 403 | |
| 404 | foreach ( array_chunk( $rcids, $this->getBatchSize() ) as $rcidBatch ) { |
| 405 | $dbw->newUpdateQueryBuilder() |
| 406 | ->update( 'recentchanges' ) |
| 407 | ->set( [ 'rc_bot' => 1 ] ) |
| 408 | ->where( [ 'rc_id' => $rcidBatch ] ) |
| 409 | ->caller( __METHOD__ )->execute(); |
| 410 | $this->waitForReplication(); |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | # Flag our recent autopatrolled edits |
| 415 | if ( !$wgMiserMode && $autopatrolgroups ) { |
| 416 | $this->output( "Flagging auto-patrolled edits...\n" ); |
| 417 | |
| 418 | $conds = [ 'rc_patrolled' => 0 ]; |
| 419 | if ( !$wgUseRCPatrol ) { |
| 420 | $subConds = []; |
| 421 | if ( $wgUseNPPatrol ) { |
| 422 | $subConds[] = $dbw->expr( 'rc_source', '=', RecentChange::SRC_NEW ); |
| 423 | } |
| 424 | if ( $wgUseFilePatrol ) { |
| 425 | $subConds[] = $dbw->expr( 'rc_log_type', '=', 'upload' ); |
| 426 | } |
| 427 | $conds[] = $dbw->makeList( $subConds, IDatabase::LIST_OR ); |
| 428 | } |
| 429 | |
| 430 | $rcids = $this->findRcIdsWithGroups( $dbw, $autopatrolgroups, $conds ); |
| 431 | foreach ( array_chunk( $rcids, $this->getBatchSize() ) as $rcidBatch ) { |
| 432 | $dbw->newUpdateQueryBuilder() |
| 433 | ->update( 'recentchanges' ) |
| 434 | ->set( [ 'rc_patrolled' => 2 ] ) |
| 435 | ->where( [ 'rc_id' => $rcidBatch ] ) |
| 436 | ->caller( __METHOD__ )->execute(); |
| 437 | $this->waitForReplication(); |
| 438 | } |
| 439 | } |
| 440 | } |
| 441 | |
| 442 | /** |
| 443 | * Rebuild pass 5: Delete duplicate entries where we generate both a page revision and a log |
| 444 | * entry for a single action (upload, move, protect, import, etc.). |
| 445 | */ |
| 446 | private function rebuildRecentChangesTablePass5() { |
| 447 | $dbw = $this->getPrimaryDB(); |
| 448 | |
| 449 | $this->output( "Removing duplicate revision and logging entries...\n" ); |
| 450 | |
| 451 | $res = $dbw->newSelectQueryBuilder() |
| 452 | ->select( [ 'ls_value', 'ls_log_id' ] ) |
| 453 | ->from( 'logging' ) |
| 454 | ->join( 'log_search', null, 'ls_log_id = log_id' ) |
| 455 | ->where( [ |
| 456 | 'ls_field' => 'associated_rev_id', |
| 457 | $dbw->expr( 'log_type', '!=', 'create' ), |
| 458 | $dbw->expr( 'log_timestamp', '>', $dbw->timestamp( $this->cutoffFrom ) ), |
| 459 | $dbw->expr( 'log_timestamp', '<', $dbw->timestamp( $this->cutoffTo ) ), |
| 460 | ] ) |
| 461 | ->caller( __METHOD__ )->fetchResultSet(); |
| 462 | |
| 463 | $updates = 0; |
| 464 | foreach ( $res as $row ) { |
| 465 | $rev_id = $row->ls_value; |
| 466 | $log_id = $row->ls_log_id; |
| 467 | |
| 468 | // Mark the logging row as having an associated rev id |
| 469 | $dbw->newUpdateQueryBuilder() |
| 470 | ->update( 'recentchanges' ) |
| 471 | ->set( [ 'rc_this_oldid' => $rev_id ] ) |
| 472 | ->where( [ 'rc_logid' => $log_id ] ) |
| 473 | ->caller( __METHOD__ )->execute(); |
| 474 | |
| 475 | // Delete the revision row |
| 476 | $dbw->newDeleteQueryBuilder() |
| 477 | ->deleteFrom( 'recentchanges' ) |
| 478 | ->where( [ 'rc_this_oldid' => $rev_id, 'rc_logid' => 0 ] ) |
| 479 | ->caller( __METHOD__ )->execute(); |
| 480 | |
| 481 | if ( ( ++$updates % $this->getBatchSize() ) == 0 ) { |
| 482 | $this->waitForReplication(); |
| 483 | } |
| 484 | } |
| 485 | } |
| 486 | |
| 487 | /** |
| 488 | * Purge cached feeds in $wanCache |
| 489 | */ |
| 490 | private function purgeFeeds() { |
| 491 | global $wgFeedClasses; |
| 492 | |
| 493 | $this->output( "Deleting feed timestamps.\n" ); |
| 494 | |
| 495 | $wanCache = $this->getServiceContainer()->getMainWANObjectCache(); |
| 496 | foreach ( $wgFeedClasses as $feed => $className ) { |
| 497 | $wanCache->delete( $wanCache->makeKey( 'rcfeed', $feed, 'timestamp' ) ); # Good enough for now. |
| 498 | } |
| 499 | } |
| 500 | } |
| 501 | |
| 502 | // @codeCoverageIgnoreStart |
| 503 | $maintClass = RebuildRecentchanges::class; |
| 504 | require_once RUN_MAINTENANCE_IF_MAIN; |
| 505 | // @codeCoverageIgnoreEnd |