fetchText.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/Maintenance.php';
 
use MediaWiki\MediaWikiServices;
use MediaWiki\Storage\BlobAccessException;
use MediaWiki\Storage\SqlBlobStore;
 
class FetchText extends Maintenance {
 
    public function __construct() {
        parent::__construct();
 
        $this->addDescription( "Fetch the raw revision blob from a blob address.\n" .
            "Integer IDs are interpreted as referring to text.old_id for backwards compatibility.\n" .
            "NOTE: Export transformations are NOT applied. " .
            "This is left to dumpTextPass.php"
        );
    }
 
    private function getBlobStore() {
        return MediaWikiServices::getInstance()->getBlobStore();
    }
 
    public function execute() {
        $stdin = $this->getStdin();
        while ( !feof( $stdin ) ) {
            $line = fgets( $stdin );
            if ( $line === false ) {
                // We appear to have lost contact...
                break;
            }
            $blobAddress = trim( $line );
 
            // Plain integers are supported for backwards compatibility with pre-MCR dumps.
            if ( strpos( $blobAddress, ':' ) === false && is_numeric( $blobAddress ) ) {
                $blobAddress = SqlBlobStore::makeAddressFromTextId( intval( $blobAddress ) );
            }
 
            try {
                $text = $this->getBlobStore()->getBlob( $blobAddress );
                $textLen = strlen( $text );
            } catch ( BlobAccessException $ex ) {
                // XXX: log $ex to stderr?
                $textLen = '-1';
                $text = '';
            } catch ( InvalidArgumentException $ex ) {
                // XXX: log $ex to stderr?
                $textLen = '-1';
                $text = '';
            }
 
            $this->output( $blobAddress . "\n" . $textLen . "\n" . $text );
        }
    }
 
}
 
$maintClass = FetchText::class;
require_once RUN_MAINTENANCE_IF_MAIN;