2022-02-16 23:29:10 +00:00
|
|
|
<?php
|
|
|
|
|
|
|
|
namespace MediaWiki\Extension\DiscussionTools\Maintenance;
|
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
use IDatabase;
|
2022-11-08 23:03:00 +00:00
|
|
|
use Language;
|
2022-09-07 01:33:37 +00:00
|
|
|
use Maintenance;
|
2022-02-16 23:29:10 +00:00
|
|
|
use MediaWiki\Extension\DiscussionTools\Hooks\HookUtils;
|
|
|
|
use MediaWiki\Extension\DiscussionTools\ThreadItemStore;
|
|
|
|
use MediaWiki\MediaWikiServices;
|
|
|
|
use MediaWiki\Revision\RevisionStore;
|
2022-09-07 01:33:37 +00:00
|
|
|
use MediaWiki\Shell\Shell;
|
2022-02-16 23:29:10 +00:00
|
|
|
use MWExceptionRenderer;
|
|
|
|
use stdClass;
|
|
|
|
use Throwable;
|
|
|
|
use Title;
|
2022-09-07 01:33:37 +00:00
|
|
|
use Wikimedia\Rdbms\SelectQueryBuilder;
|
2022-02-16 23:29:10 +00:00
|
|
|
|
2022-09-29 17:00:38 +00:00
|
|
|
$IP = getenv( 'MW_INSTALL_PATH' );
|
|
|
|
if ( $IP === false ) {
|
|
|
|
$IP = __DIR__ . '/../../..';
|
2022-02-16 23:29:10 +00:00
|
|
|
}
|
2022-09-29 17:00:38 +00:00
|
|
|
require_once "$IP/maintenance/Maintenance.php";
|
2022-02-16 23:29:10 +00:00
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
class PersistRevisionThreadItems extends Maintenance {
|
|
|
|
|
2022-10-21 19:34:18 +00:00
|
|
|
private IDatabase $dbr;
|
|
|
|
private ThreadItemStore $itemStore;
|
|
|
|
private RevisionStore $revStore;
|
2022-11-08 23:03:00 +00:00
|
|
|
private Language $lang;
|
2022-02-16 23:29:10 +00:00
|
|
|
|
|
|
|
public function __construct() {
|
|
|
|
parent::__construct();
|
|
|
|
$this->requireExtension( 'DiscussionTools' );
|
|
|
|
$this->addDescription( 'Persist thread item information for the given pages/revisions' );
|
|
|
|
$this->addOption( 'rev', 'Revision ID to process', false, true, false, true );
|
|
|
|
$this->addOption( 'page', 'Page title to process', false, true, false, true );
|
2022-09-07 01:33:37 +00:00
|
|
|
$this->addOption( 'all', 'Process the whole wiki' );
|
|
|
|
$this->addOption( 'current', 'Process current revisions only' );
|
|
|
|
$this->addOption( 'start', 'Restart from this position (as printed by the script)', false, true );
|
|
|
|
$this->setBatchSize( 100 );
|
2022-02-16 23:29:10 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
public function execute() {
|
|
|
|
$services = MediaWikiServices::getInstance();
|
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
$this->dbr = $this->getDB( DB_REPLICA );
|
2022-02-16 23:29:10 +00:00
|
|
|
$this->itemStore = $services->getService( 'DiscussionTools.ThreadItemStore' );
|
|
|
|
$this->revStore = $services->getRevisionStore();
|
2022-11-08 23:03:00 +00:00
|
|
|
$this->lang = $services->getLanguageFactory()->getLanguage( 'en' );
|
2022-02-16 23:29:10 +00:00
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
$qb = $this->dbr->newSelectQueryBuilder();
|
|
|
|
|
|
|
|
$qb->queryInfo( $this->revStore->getQueryInfo( [ 'page' ] ) );
|
|
|
|
|
2022-02-16 23:29:10 +00:00
|
|
|
if ( $this->getOption( 'all' ) ) {
|
2022-09-07 01:33:37 +00:00
|
|
|
// Do nothing
|
2022-02-16 23:29:10 +00:00
|
|
|
|
|
|
|
} elseif ( $this->getOption( 'page' ) ) {
|
|
|
|
$linkBatch = $services->getLinkBatchFactory()->newLinkBatch();
|
|
|
|
foreach ( $this->getOption( 'page' ) as $page ) {
|
|
|
|
$linkBatch->addObj( Title::newFromText( $page ) );
|
|
|
|
}
|
|
|
|
$pageIds = array_map( static function ( $page ) {
|
|
|
|
return $page->getId();
|
|
|
|
}, $linkBatch->getPageIdentities() );
|
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
$qb->where( [ 'rev_page' => $pageIds ] );
|
2022-02-16 23:29:10 +00:00
|
|
|
|
|
|
|
} elseif ( $this->getOption( 'rev' ) ) {
|
2022-09-07 01:33:37 +00:00
|
|
|
$qb->where( [ 'rev_id' => $this->getOption( 'rev' ) ] );
|
2022-02-16 23:29:10 +00:00
|
|
|
} else {
|
|
|
|
$this->error( "One of 'all', 'page', or 'rev' required" );
|
|
|
|
$this->maybeHelp( true );
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2022-08-22 23:38:17 +00:00
|
|
|
if ( $this->getOption( 'current' ) ) {
|
2022-09-07 01:33:37 +00:00
|
|
|
$qb->where( 'rev_id = page_latest' );
|
|
|
|
$index = [ 'page_id' ];
|
|
|
|
} else {
|
|
|
|
// Process in order by page and time to avoid confusing results while the script is running
|
|
|
|
$index = [ 'rev_page', 'rev_timestamp', 'rev_id' ];
|
|
|
|
}
|
|
|
|
|
|
|
|
$this->process( $qb, $index );
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* @param SelectQueryBuilder $qb
|
|
|
|
* @param array $index
|
|
|
|
*/
|
|
|
|
private function process( SelectQueryBuilder $qb, array $index ): void {
|
2022-11-08 23:03:00 +00:00
|
|
|
$start = microtime( true );
|
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
$qb->caller( __METHOD__ );
|
|
|
|
|
|
|
|
// estimateRowCount() refuses to work when fields are set, so we can't just call it on $qb
|
|
|
|
$countQueryInfo = $qb->getQueryInfo();
|
|
|
|
$count = $qb->newSubquery()
|
|
|
|
->rawTables( $countQueryInfo['tables'] )
|
|
|
|
->where( $countQueryInfo['conds'] )
|
|
|
|
->options( $countQueryInfo['options'] )
|
|
|
|
->joinConds( $countQueryInfo['join_conds'] )
|
|
|
|
->caller( __METHOD__ )
|
|
|
|
->estimateRowCount();
|
|
|
|
$this->output( "Processing... (estimated $count rows)\n" );
|
|
|
|
|
|
|
|
$processed = 0;
|
|
|
|
$updated = 0;
|
|
|
|
|
|
|
|
$qb->orderBy( $index );
|
|
|
|
$batchSize = $this->getBatchSize();
|
|
|
|
$qb->limit( $batchSize );
|
|
|
|
|
|
|
|
$batchStart = null;
|
|
|
|
if ( $this->getOption( 'start' ) ) {
|
|
|
|
$batchStart = json_decode( $this->getOption( 'start' ) );
|
|
|
|
if ( !$batchStart ) {
|
|
|
|
$this->error( "Invalid 'start'" );
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
while ( true ) {
|
|
|
|
$qbForBatch = clone $qb;
|
|
|
|
if ( $batchStart ) {
|
|
|
|
$batchStartCond = $this->dbr->buildComparison( '>', array_combine( $index, $batchStart ) );
|
|
|
|
$qbForBatch->where( $batchStartCond );
|
|
|
|
|
|
|
|
$batchStartOutput = Shell::escape( json_encode( $batchStart ) );
|
|
|
|
$this->output( "--start $batchStartOutput\n" );
|
|
|
|
}
|
|
|
|
|
|
|
|
$res = $qbForBatch->fetchResultSet();
|
|
|
|
foreach ( $res as $row ) {
|
|
|
|
$updated += (int)$this->processRow( $row );
|
|
|
|
}
|
|
|
|
$processed += $res->numRows();
|
|
|
|
|
|
|
|
$this->output( "Processed $processed (updated $updated) of $count rows\n" );
|
|
|
|
|
|
|
|
$this->waitForReplication();
|
|
|
|
|
|
|
|
if ( $res->numRows() < $batchSize || !isset( $row ) ) {
|
|
|
|
// Done
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Update the conditions to select the next batch.
|
|
|
|
$batchStart = [];
|
|
|
|
foreach ( $index as $field ) {
|
|
|
|
$batchStart[] = $row->$field;
|
|
|
|
}
|
2022-08-22 23:38:17 +00:00
|
|
|
}
|
|
|
|
|
2022-11-08 23:03:00 +00:00
|
|
|
$duration = microtime( true ) - $start;
|
|
|
|
$durationFormatted = $this->lang->formatTimePeriod( $duration );
|
|
|
|
$this->output( "Finished in $durationFormatted\n" );
|
2022-02-16 23:29:10 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* @param stdClass $row Database table row
|
2022-09-07 01:33:37 +00:00
|
|
|
* @return bool
|
2022-02-16 23:29:10 +00:00
|
|
|
*/
|
2022-09-07 01:33:37 +00:00
|
|
|
private function processRow( stdClass $row ): bool {
|
2022-02-16 23:29:10 +00:00
|
|
|
$changed = false;
|
|
|
|
try {
|
|
|
|
$rev = $this->revStore->newRevisionFromRow( $row );
|
|
|
|
$title = Title::newFromLinkTarget(
|
|
|
|
$rev->getPageAsLinkTarget()
|
|
|
|
);
|
|
|
|
if ( HookUtils::isAvailableForTitle( $title ) ) {
|
2022-12-13 15:50:02 +00:00
|
|
|
$threadItemSet = HookUtils::parseRevisionParsoidHtml( $rev, false );
|
2022-02-16 23:29:10 +00:00
|
|
|
|
2022-09-07 01:33:37 +00:00
|
|
|
// Store permalink data
|
|
|
|
$changed = $this->itemStore->insertThreadItems( $rev, $threadItemSet );
|
2022-02-16 23:29:10 +00:00
|
|
|
}
|
|
|
|
} catch ( Throwable $e ) {
|
|
|
|
$this->output( "Error while processing revid=$row->rev_id, pageid=$row->rev_page\n" );
|
|
|
|
MWExceptionRenderer::output( $e, MWExceptionRenderer::AS_RAW );
|
|
|
|
}
|
2022-09-07 01:33:37 +00:00
|
|
|
return $changed;
|
2022-02-16 23:29:10 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
$maintClass = PersistRevisionThreadItems::class;
|
|
|
|
require_once RUN_MAINTENANCE_IF_MAIN;
|