dumpIterator.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
abstract class DumpIterator extends Maintenance {
 
    private $count = 0;
    private $startTime;
 
    public function __construct() {
        parent::__construct();
        $this->mDescription = "Does something with a dump";
        $this->addOption( 'file', 'File with text to run.', false, true );
        $this->addOption( 'dump', 'XML dump to execute all revisions.', false, true );
        $this->addOption( 'from', 'Article from XML dump to start from.', false, true );
    }
 
    public function execute() {
        if ( !( $this->hasOption( 'file' ) ^ $this->hasOption( 'dump' ) ) ) {
            $this->error( "You must provide a file or dump", true );
        }
 
        $this->checkOptions();
 
        if ( $this->hasOption( 'file' ) ) {
            $revision = new WikiRevision;
 
            $revision->setText( file_get_contents( $this->getOption( 'file' ) ) );
            $revision->setTitle( Title::newFromText( rawurldecode( basename( $this->getOption( 'file' ), '.txt' ) ) ) );
            $this->handleRevision( $revision );
            return;
        }
 
        $this->startTime = microtime( true );
 
        if ( $this->getOption( 'dump' ) == '-' ) {
            $source = new ImportStreamSource( $this->getStdin() );
        } else {
            $this->error( "Sorry, I don't support dump filenames yet. Use - and provide it on stdin on the meantime.", true );
        }
        $importer = new WikiImporter( $source );
 
        $importer->setRevisionCallback(
            array( &$this, 'handleRevision' ) );
 
        $this->from = $this->getOption( 'from', null );
        $this->count = 0;
        $importer->doImport();
 
        $this->conclusions();
 
        $delta = microtime( true ) - $this->startTime;
        $this->error( "Done {$this->count} revisions in " . round( $delta, 2 ) . " seconds " );
        if ( $delta > 0 ) {
            $this->error( round( $this->count / $delta, 2 ) . " pages/sec" );
        }
 
        # Perform the memory_get_peak_usage() when all the other data has been output so there's no damage if it dies.
        # It is only available since 5.2.0 (since 5.2.1 if you haven't compiled with --enable-memory-limit)
        $this->error( "Memory peak usage of " . memory_get_peak_usage() . " bytes\n" );
    }
 
    public function finalSetup() {
        parent::finalSetup();
 
        if ( $this->getDbType() == Maintenance::DB_NONE ) {
            global $wgUseDatabaseMessages, $wgLocalisationCacheConf, $wgHooks;
            $wgUseDatabaseMessages = false;
            $wgLocalisationCacheConf['storeClass'] = 'LCStoreNull';
            $wgHooks['InterwikiLoadPrefix'][] = 'DumpIterator::disableInterwikis';
        }
    }
 
    static function disableInterwikis( $prefix, &$data ) {
        # Title::newFromText will check on each namespaced article if it's an interwiki.
        # We always answer that it is not.
 
        return false;
    }
 
    public function handleRevision( $rev ) {
        $title = $rev->getTitle();
        if ( !$title ) {
            $this->error( "Got bogus revision with null title!" );
            return;
        }
 
        $this->count++;
        if ( isset( $this->from ) ) {
            if ( $this->from != $title ) {
                return;
            }
            $this->output( "Skipped " . ( $this->count - 1 ) . " pages\n" );
 
            $this->count = 1;
            $this->from = null;
        }
 
        $this->processRevision( $rev );
    }
 
    /* Stub function for processing additional options */
    public function checkOptions() {
        return;
    }
 
    /* Stub function for giving data about what was computed */
    public function conclusions() {
        return;
    }
 
    /* Core function which does whatever the maintenance script is designed to do */
    abstract public function processRevision( $rev );
}
 
class SearchDump extends DumpIterator {
 
    public function __construct() {
        parent::__construct();
        $this->mDescription = "Runs a regex in the revisions from a dump";
        $this->addOption( 'regex', 'Searching regex', true, true );
    }
 
    public function getDbType() {
        return Maintenance::DB_NONE;
    }
 
    public function processRevision( $rev ) {
        if ( preg_match( $this->getOption( 'regex' ), $rev->getContent()->getTextForSearchIndex() ) ) {
            $this->output( $rev->getTitle() . " matches at edit from " . $rev->getTimestamp() . "\n" );
        }
    }
}
 
$maintClass = "SearchDump";
require_once RUN_MAINTENANCE_IF_MAIN;