TextPassDumper.php

Go to the documentation of this file.

<?php
namespace MediaWiki\Maintenance;
 
// @codeCoverageIgnoreStart
require_once __DIR__ . '/BackupDumper.php';
require_once __DIR__ . '/../../includes/Export/WikiExporter.php';
// @codeCoverageIgnoreEnd
 
use Exception;
use MediaWiki\Content\UnknownContentModelException;
use MediaWiki\Exception\MWException;
use MediaWiki\Export\BaseDump;
use MediaWiki\Export\ExportProgressFilter;
use MediaWiki\Export\WikiExporter;
use MediaWiki\Export\XmlDumpWriter;
use MediaWiki\Revision\RevisionStore;
use MediaWiki\Revision\SlotRecord;
use MediaWiki\Settings\SettingsBuilder;
use MediaWiki\Shell\Shell;
use MediaWiki\Storage\BlobAccessException;
use MediaWiki\Storage\BlobStore;
use MediaWiki\Storage\SqlBlobStore;
use MediaWiki\WikiMap\WikiMap;
use MediaWiki\Xml\Xml;
use RuntimeException;
use Wikimedia\Timestamp\ConvertibleTimestamp;
use Wikimedia\Timestamp\TimestampFormat as TS;
use XMLParser;
 
class TextPassDumper extends BackupDumper {
    public $prefetch = null;
    private $thisPage;
    private $thisRev;
    private $thisRole = null;
 
    public $maxTimeAllowed = 0;
 
    protected $input = "php://stdin";
    protected $history = WikiExporter::FULL;
    protected $fetchCount = 0;
    protected $prefetchCount = 0;
    protected $prefetchCountLast = 0;
    protected $fetchCountLast = 0;
 
    protected $maxFailures = 5;
    protected $maxConsecutiveFailedTextRetrievals = 200;
    protected $failureTimeout = 5;
 
    protected $bufferSize = 524_288;
 
    protected $php = [ PHP_BINARY ];
    protected $spawn = false;
 
    protected $spawnProc = false;
 
    protected $spawnWrite;
 
    protected $spawnRead;
 
    protected $spawnErr = false;
 
    protected $xmlwriterobj = false;
 
    protected $timeExceeded = false;
    protected $firstPageWritten = false;
    protected $lastPageWritten = false;
    protected $checkpointJustWritten = false;
    protected $checkpointFiles = [];
 
    public function __construct( $args = null ) {
        parent::__construct();
 
        $this->addDescription( <<<TEXT
This script postprocesses XML dumps from dumpBackup.php to add
page text which was stubbed out (using --stub).
 
XML input is accepted on stdin.
XML output is sent to stdout; progress reports are sent to stderr.
TEXT
        );
        $this->stderr = fopen( "php://stderr", "wt" );
 
        $this->addOption( 'stub', 'To load a compressed stub dump instead of stdin. ' .
            'Specify as --stub=<type>:<file>.', false, true );
        $this->addOption( 'prefetch', 'Use a prior dump file as a text source, to savepressure on the ' .
            'database. (Requires the XMLReader extension). Specify as --prefetch=<type>:<file>',
            false, true );
        $this->addOption( 'maxtime', 'Write out checkpoint file after this many minutes (writing' .
            'out complete page, closing xml file properly, and opening new one' .
            'with header).  This option requires the checkpointfile option.', false, true );
        $this->addOption( 'checkpointfile', 'Use this string for checkpoint filenames,substituting ' .
            'first pageid written for the first %s (required) and the last pageid written for the ' .
            'second %s if it exists.', false, true, false, true ); // This can be specified multiple times
        $this->addOption( 'quiet', 'Don\'t dump status reports to stderr.' );
        $this->addOption( 'full', 'Dump all revisions of every page' );
        $this->addOption( 'current', 'Base ETA on number of pages in database instead of all revisions' );
        $this->addOption( 'spawn', 'Spawn a subprocess for loading text records, optionally specify ' .
            'php[,mwscript] paths' );
        $this->addOption( 'buffersize', 'Buffer size in bytes to use for reading the stub. ' .
            '(Default: 512 KiB, Minimum: 4 KiB)', false, true );
 
        if ( $args ) {
            $this->loadWithArgv( $args );
            $this->processOptions();
        }
    }
 
    public function finalSetup( SettingsBuilder $settingsBuilder ) {
        parent::finalSetup( $settingsBuilder );
 
        SevenZipStream::register();
    }
 
    private function getBlobStore() {
        return $this->getServiceContainer()->getBlobStore();
    }
 
    private function getRevisionStore() {
        return $this->getServiceContainer()->getRevisionStore();
    }
 
    public function execute() {
        $this->processOptions();
        $this->dump( $this->history );
    }
 
    protected function processOptions() {
        parent::processOptions();
 
        if ( $this->hasOption( 'buffersize' ) ) {
            $this->bufferSize = max( intval( $this->getOption( 'buffersize' ) ), 4 * 1024 );
        }
 
        if ( $this->hasOption( 'prefetch' ) ) {
            $url = $this->processFileOpt( $this->getOption( 'prefetch' ) );
            $this->prefetch = new BaseDump( $url );
        }
 
        if ( $this->hasOption( 'stub' ) ) {
            $this->input = $this->processFileOpt( $this->getOption( 'stub' ) );
        }
 
        if ( $this->hasOption( 'maxtime' ) ) {
            $this->maxTimeAllowed = intval( $this->getOption( 'maxtime' ) ) * 60;
        }
 
        if ( $this->hasOption( 'checkpointfile' ) ) {
            $this->checkpointFiles = $this->getOption( 'checkpointfile' );
        }
 
        if ( $this->hasOption( 'current' ) ) {
            $this->history = WikiExporter::CURRENT;
        }
 
        if ( $this->hasOption( 'full' ) ) {
            $this->history = WikiExporter::FULL;
        }
 
        if ( $this->hasOption( 'spawn' ) ) {
            $this->spawn = true;
            $val = $this->getOption( 'spawn' );
            if ( $val !== 1 ) {
                $this->php = explode( ',', $val, 2 );
            }
        }
    }
 
    public function initProgress( $history = WikiExporter::FULL ) {
        parent::initProgress( $history );
        $this->timeOfCheckpoint = $this->startTime;
    }
 
    public function dump( $history, $text = WikiExporter::TEXT ) {
        // Notice messages will foul up your XML output even if they're
        // relatively harmless.
        if ( ini_get( 'display_errors' ) ) {
            ini_set( 'display_errors', 'stderr' );
        }
 
        $this->initProgress( $history );
 
        $this->egress = new ExportProgressFilter( $this->sink, $this );
 
        // it would be nice to do it in the constructor, oh well. need egress set
        $this->finalOptionCheck();
 
        // we only want this so we know how to close a stream :-P
        $this->xmlwriterobj = new XmlDumpWriter( XmlDumpWriter::WRITE_CONTENT, $this->schemaVersion );
 
        $input = fopen( $this->input, "rt" );
        $this->readDump( $input );
 
        if ( $this->spawnProc ) {
            $this->closeSpawn();
        }
 
        $this->report( true );
    }
 
    protected function processFileOpt( string $opt ): string {
        $split = explode( ':', $opt, 2 );
        $val = $split[0];
        $param = $split[1] ?? '';
        $newFileURIs = [];
        foreach ( explode( ';', $param ) as $uri ) {
            $newFileURIs[] = match ( $val ) {
                'gzip' => "compress.zlib://$uri",
                'bzip2' => "compress.bzip2://$uri",
                '7zip' => "mediawiki.compress.7z://$uri",
                default => $uri,
            };
        }
        return implode( ';', $newFileURIs );
    }
 
    public function showReport() {
        if ( !$this->prefetch ) {
            parent::showReport();
 
            return;
        }
 
        if ( $this->reporting ) {
            $now = ConvertibleTimestamp::now( TS::DB );
            $nowts = microtime( true );
            $deltaAll = $nowts - $this->startTime;
            $deltaPart = $nowts - $this->lastTime;
            $this->pageCountPart = $this->pageCount - $this->pageCountLast;
            $this->revCountPart = $this->revCount - $this->revCountLast;
 
            if ( $deltaAll ) {
                $portion = $this->revCount / $this->maxCount;
                $eta = $this->startTime + $deltaAll / $portion;
                $etats = wfTimestamp( TS::DB, intval( $eta ) );
                if ( $this->fetchCount ) {
                    $fetchRate = 100.0 * $this->prefetchCount / $this->fetchCount;
                } else {
                    $fetchRate = '-';
                }
                $pageRate = $this->pageCount / $deltaAll;
                $revRate = $this->revCount / $deltaAll;
            } else {
                $pageRate = '-';
                $revRate = '-';
                $etats = '-';
                $fetchRate = '-';
            }
            if ( $deltaPart ) {
                if ( $this->fetchCountLast ) {
                    $fetchRatePart = 100.0 * $this->prefetchCountLast / $this->fetchCountLast;
                } else {
                    $fetchRatePart = '-';
                }
                $pageRatePart = $this->pageCountPart / $deltaPart;
                $revRatePart = $this->revCountPart / $deltaPart;
            } else {
                $fetchRatePart = '-';
                $pageRatePart = '-';
                $revRatePart = '-';
            }
 
            $dbDomain = WikiMap::getCurrentWikiDbDomain()->getId();
            $this->progress( sprintf(
                "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
                    . "%d revs (%0.1f|%0.1f/sec all|curr), %0.1f%%|%0.1f%% "
                    . "prefetched (all|curr), ETA %s [max %d]",
                $now, $dbDomain, $this->ID, $this->pageCount, $pageRate,
                $pageRatePart, $this->revCount, $revRate, $revRatePart,
                $fetchRate, $fetchRatePart, $etats, $this->maxCount
            ) );
            $this->lastTime = $nowts;
            $this->revCountLast = $this->revCount;
            $this->prefetchCountLast = $this->prefetchCount;
            $this->fetchCountLast = $this->fetchCount;
        }
    }
 
    private function setTimeExceeded() {
        $this->timeExceeded = true;
    }
 
    private function checkIfTimeExceeded(): bool {
        if ( $this->maxTimeAllowed
            && ( $this->lastTime - $this->timeOfCheckpoint > $this->maxTimeAllowed )
        ) {
            return true;
        }
 
        return false;
    }
 
    private function finalOptionCheck() {
        if ( ( $this->checkpointFiles && !$this->maxTimeAllowed )
            || ( $this->maxTimeAllowed && !$this->checkpointFiles )
        ) {
            throw new RuntimeException( "Options checkpointfile and maxtime must be specified together.\n" );
        }
        foreach ( $this->checkpointFiles as $checkpointFile ) {
            $count = substr_count( $checkpointFile, "%s" );
            if ( $count !== 2 ) {
                throw new RuntimeException( "Option checkpointfile must contain two '%s' "
                    . "for substitution of first and last pageids, count is $count instead, "
                    . "file is $checkpointFile.\n" );
            }
        }
 
        if ( $this->checkpointFiles ) {
            $filenameList = (array)$this->egress->getFilenames();
            if ( count( $filenameList ) !== count( $this->checkpointFiles ) ) {
                throw new RuntimeException( "One checkpointfile must be specified "
                    . "for each output option, if maxtime is used.\n" );
            }
        }
    }
 
    protected function readDump( $input ) {
        $this->buffer = "";
        $this->openElement = false;
        $this->atStart = true;
        $this->state = "";
        $this->lastName = "";
        $this->thisPage = "";
        $this->thisRev = "";
        $this->thisRole = null;
        $this->thisRevModel = null;
        $this->thisRevFormat = null;
 
        $parser = xml_parser_create( "UTF-8" );
        xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, 0 );
 
        xml_set_element_handler(
            $parser,
            $this->startElement( ... ),
            $this->endElement( ... )
        );
        xml_set_character_data_handler( $parser, $this->characterData( ... ) );
 
        $offset = 0; // for context extraction on error reporting
        do {
            if ( $this->checkIfTimeExceeded() ) {
                $this->setTimeExceeded();
            }
            $chunk = fread( $input, $this->bufferSize );
            if ( !xml_parse( $parser, $chunk, feof( $input ) ) ) {
                wfDebug( "TextDumpPass::readDump encountered XML parsing error" );
 
                $byte = xml_get_current_byte_index( $parser );
                $msg = wfMessage( 'xml-error-string',
                    'XML import parse failure',
                    xml_get_current_line_number( $parser ),
                    xml_get_current_column_number( $parser ),
                    $byte . ( $chunk === false ? '' : ( '; "' . substr( $chunk, $byte - $offset, 16 ) . '"' ) ),
                    xml_error_string( xml_get_error_code( $parser ) ) )->escaped();
 
                throw new MWException( $msg );
            }
            $offset += strlen( $chunk );
        } while ( $chunk !== false && !feof( $input ) );
        if ( $this->maxTimeAllowed ) {
            $filenameList = (array)$this->egress->getFilenames();
            // we wrote some stuff after last checkpoint that needs renamed
            if ( file_exists( $filenameList[0] ) ) {
                $newFilenames = [];
                # we might have just written the header and footer and had no
                # pages or revisions written... perhaps they were all deleted
                # there's no pageID 0 so we use that. the caller is responsible
                # for deciding what to do with a file containing only the
                # siteinfo information and the mw tags.
                if ( !$this->firstPageWritten ) {
                    $firstPageID = str_pad( '0', 9, "0", STR_PAD_LEFT );
                    $lastPageID = str_pad( '0', 9, "0", STR_PAD_LEFT );
                } else {
                    $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
                    $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
                }
 
                $filenameCount = count( $filenameList );
                for ( $i = 0; $i < $filenameCount; $i++ ) {
                    $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
                    $fileinfo = pathinfo( $filenameList[$i] );
                    $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
                }
                $this->egress->closeAndRename( $newFilenames );
            }
        }
 
        return true;
    }
 
    private function exportTransform( $text, $model, $format = null ) {
        try {
            $contentHandler = $this->getServiceContainer()
                ->getContentHandlerFactory()
                ->getContentHandler( $model );
        } catch ( UnknownContentModelException $ex ) {
            wfWarn( "Unable to apply export transformation for content model '$model': " .
                $ex->getMessage() );
 
            $this->progress(
                "Unable to apply export transformation for content model '$model': " .
                $ex->getMessage()
            );
            return $text;
        }
 
        return $contentHandler->exportTransform( $text, $format );
    }
 
    protected function getText( $id, $model = null, $format = null, $expSize = null ) {
        if ( !$this->isValidTextId( $id ) ) {
            $msg = "Skipping bad text id " . $id . " of revision " . $this->thisRev;
            $this->progress( $msg );
            return '';
        }
 
        $model = $model ?: null;
        $prefetchNotTried = true; // Whether or not we already tried to get the text via prefetch.
        $text = false; // The candidate for a good text. false if no proper value.
        $failures = 0; // The number of times, this invocation of getText already failed.
        $contentAddress = $id; // Where the content should be found
 
        // The number of times getText failed without yielding a good text in between.
        static $consecutiveFailedTextRetrievals = 0;
 
        $this->fetchCount++;
 
        // To allow to simply return on success and do not have to worry about book keeping,
        // we assume, this fetch works (possible after some retries). Nevertheless, we keep
        // the old value, so we can restore it, if problems occur (See after the while loop).
        $oldConsecutiveFailedTextRetrievals = $consecutiveFailedTextRetrievals;
        $consecutiveFailedTextRetrievals = 0;
 
        while ( $failures < $this->maxFailures ) {
            // As soon as we found a good text for the $id, we will return immediately.
            // Hence, if we make it past the try catch block, we know that we did not
            // find a good text.
 
            try {
                // Step 1: Get some text (or reuse from previous iteratuon if checking
                //         for plausibility failed)
 
                // Trying to get prefetch, if it has not been tried before
                // @phan-suppress-next-line PhanRedundantValueComparisonInLoop
                if ( $text === false && $this->prefetch && $prefetchNotTried ) {
                    $prefetchNotTried = false;
                    $tryIsPrefetch = true;
                    $text = $this->prefetch->prefetch(
                        (int)$this->thisPage,
                        (int)$this->thisRev,
                        trim( $this->thisRole )
                    ) ?? false;
 
                    if ( is_string( $text ) && $model !== null ) {
                        // Apply export transformation to text coming from an old dump.
                        // The purpose of this transformation is to convert up from legacy
                        // formats, which may still be used in the older dump that is used
                        // for pre-fetching. Applying the transformation again should not
                        // interfere with content that is already in the correct form.
                        $text = $this->exportTransform( $text, $model, $format );
                    }
                }
 
                if ( $text === false ) {
                    // Fallback to asking the database
                    $tryIsPrefetch = false;
                    if ( $this->spawn ) {
                        $text = $this->getTextSpawned( $contentAddress );
                    } else {
                        $text = $this->getTextDb( $contentAddress );
                    }
 
                    if ( $text !== false && $model !== null ) {
                        // Apply export transformation to text coming from the database.
                        // Prefetched text should already have transformations applied.
                        $text = $this->exportTransform( $text, $model, $format );
                    }
 
                    // No more checks for texts from DB for now.
                    // If we received something that is not false,
                    // We treat it as good text, regardless of whether it actually is or is not
                    if ( $text !== false ) {
                        return $text;
                    }
                }
 
                if ( $text === false ) {
                    throw new RuntimeException( "Generic error while obtaining text for id " . $contentAddress );
                }
 
                // We received a good candidate for the text of $id via some method
 
                // Step 2: Checking for plausibility and return the text if it is
                //         plausible
 
                if ( $expSize === null || strlen( $text ) == $expSize ) {
                    // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Set when text is not false
                    if ( $tryIsPrefetch ) {
                        $this->prefetchCount++;
                    }
 
                    return $text;
                }
 
                $text = false;
                throw new RuntimeException( "Received text is unplausible for id " . $contentAddress );
            } catch ( Exception $e ) {
                $msg = "getting/checking text " . $contentAddress . " failed (" . $e->getMessage()
                    . ") for revision " . $this->thisRev;
                if ( $failures + 1 < $this->maxFailures ) {
                    $msg .= " (Will retry " . ( $this->maxFailures - $failures - 1 ) . " more times)";
                }
                $this->progress( $msg );
            }
 
            // Something went wrong; we did not get a text that was plausible :(
            $failures++;
 
            if ( $contentAddress === $id && $this->thisRev && trim( $this->thisRole ) ) {
                try {
                    // MediaWiki doesn't guarantee that content addresses are valid
                    // for any significant length of time. Try refreshing as the
                    // previously retrieved address may no longer be valid.
                    $revRecord = $this->getRevisionStore()->getRevisionById( (int)$this->thisRev );
                    if ( $revRecord !== null ) {
                        $refreshed = $revRecord->getSlot( trim( $this->thisRole ) )->getAddress();
                        if ( $contentAddress !== $refreshed ) {
                            $this->progress(
                                "Updated content address for rev {$this->thisRev} from "
                                . "{$contentAddress} to {$refreshed}"
                            );
                            $contentAddress = $refreshed;
                            // Skip sleeping if we updated the address
                            continue;
                        }
                    }
                } catch ( Exception $e ) {
                    $this->progress(
                        "refreshing content address for revision {$this->thisRev} failed ({$e->getMessage()})"
                    );
                }
            }
 
            // A failure in a prefetch hit does not warrant resetting db connection etc.
            // @phan-suppress-next-line PhanPossiblyUndeclaredVariable Set in the prefetch block above
            if ( !$tryIsPrefetch ) {
                // After backing off for some time, we try to reboot the whole process as
                // much as possible to not carry over failures from one part to the other
                // parts
                sleep( $this->failureTimeout );
                try {
                    if ( $this->spawn ) {
                        $this->closeSpawn();
                        $this->openSpawn();
                    }
                } catch ( Exception $e ) {
                    $this->progress( "Rebooting getText infrastructure failed (" . $e->getMessage() . ")" .
                        " Trying to continue anyways" );
                }
            }
        }
 
        // Retrieving a good text for $id failed (at least) maxFailures times.
        // We abort for this $id.
 
        // Restoring the consecutive failures, and maybe aborting, if the dump
        // is too broken.
        $consecutiveFailedTextRetrievals = $oldConsecutiveFailedTextRetrievals + 1;
        if ( $consecutiveFailedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals ) {
            throw new MWException( "Graceful storage failure" );
        }
 
        return "";
    }
 
    private function getTextDb( $id ) {
        $store = $this->getBlobStore();
        $address = ( is_int( $id ) || !str_contains( $id, ':' ) )
            ? SqlBlobStore::makeAddressFromTextId( (int)$id )
            : $id;
 
        try {
            $text = $store->getBlob( $address );
 
            $stripped = str_replace( "\r", "", $text );
            $normalized = $this->getServiceContainer()->getContentLanguage()
                ->normalize( $stripped );
 
            return $normalized;
        } catch ( BlobAccessException ) {
            // XXX: log a warning?
            return false;
        }
    }
 
    private function getTextSpawned( $address ) {
        if ( !$this->spawnProc ) {
            // First time?
            // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
            @$this->openSpawn();
        }
        // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
        return @$this->getTextSpawnedOnce( $address );
    }
 
    protected function openSpawn(): bool {
        $wiki = WikiMap::getCurrentWikiId();
        if ( count( $this->php ) == 2 ) {
            $mwscriptpath = $this->php[1];
        } else {
            // FIXME: Avoid this hardcoded wmf-config reference.
            // Perhaps refactor the below by using wfShellWikiCmd or use the
            // 'wrapper' option which is already injected for this purpose.
            $mwscriptpath = MW_INSTALL_PATH . '/../multiversion/MWScript.php';
        }
        if ( file_exists( $mwscriptpath ) ) {
            $cmd = implode( " ",
                array_map( Shell::escape( ... ),
                    [
                        $this->php[0],
                        $mwscriptpath,
                        "fetchText.php",
                        '--wiki', $wiki ] ) );
        } else {
            $cmd = implode( " ",
                array_map( Shell::escape( ... ),
                    [
                        $this->php[0],
                        MW_INSTALL_PATH . '/maintenance/fetchText.php',
                        '--wiki', $wiki ] ) );
        }
        $spec = [
            0 => [ "pipe", "r" ],
            1 => [ "pipe", "w" ],
            2 => [ "file", "/dev/null", "a" ] ];
        $pipes = [];
 
        $this->progress( "Spawning database subprocess: $cmd" );
        $this->spawnProc = proc_open( $cmd, $spec, $pipes );
        if ( !$this->spawnProc ) {
            $this->progress( "Subprocess spawn failed." );
 
            return false;
        }
        [
            $this->spawnWrite, // -> stdin
            $this->spawnRead, // <- stdout
        ] = $pipes;
 
        return true;
    }
 
    private function closeSpawn() {
        if ( $this->spawnRead ) {
            // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
            @fclose( $this->spawnRead );
        }
        $this->spawnRead = null;
        if ( $this->spawnWrite ) {
            // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
            @fclose( $this->spawnWrite );
        }
        $this->spawnWrite = null;
        if ( $this->spawnErr ) {
            // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
            @fclose( $this->spawnErr );
        }
        $this->spawnErr = false;
        if ( $this->spawnProc ) {
            // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged
            @proc_close( $this->spawnProc );
        }
        $this->spawnProc = false;
    }
 
    private function getTextSpawnedOnce( $address ) {
        if ( is_int( $address ) || intval( $address ) ) {
            $address = SqlBlobStore::makeAddressFromTextId( (int)$address );
        }
 
        $ok = fwrite( $this->spawnWrite, "$address\n" );
        // $this->progress( ">> $id" );
        if ( !$ok ) {
            return false;
        }
 
        $ok = fflush( $this->spawnWrite );
        // $this->progress( ">> [flush]" );
        if ( !$ok ) {
            return false;
        }
 
        // check that the text address they are sending is the one we asked for
        // this avoids out of sync revision text errors we have encountered in the past
        $newAddress = fgets( $this->spawnRead );
        if ( $newAddress === false ) {
            return false;
        }
        $newAddress = trim( $newAddress );
        if ( !str_contains( $newAddress, ':' ) ) {
            $newAddress = SqlBlobStore::makeAddressFromTextId( intval( $newAddress ) );
        }
 
        if ( $newAddress !== $address ) {
            return false;
        }
 
        $len = fgets( $this->spawnRead );
        // $this->progress( "<< " . trim( $len ) );
        if ( $len === false ) {
            return false;
        }
 
        $nbytes = intval( $len );
        // actual error, not zero-length text
        if ( $nbytes < 0 ) {
            return false;
        }
 
        $text = "";
 
        // Subprocess may not send everything at once, we have to loop.
        while ( $nbytes > strlen( $text ) ) {
            $buffer = fread( $this->spawnRead, $nbytes - strlen( $text ) );
            if ( $buffer === false ) {
                break;
            }
            $text .= $buffer;
        }
 
        $gotbytes = strlen( $text );
        if ( $gotbytes != $nbytes ) {
            $this->progress( "Expected $nbytes bytes from database subprocess, got $gotbytes " );
 
            return false;
        }
 
        // Do normalization in the dump thread...
        $stripped = str_replace( "\r", "", $text );
        $normalized = $this->getServiceContainer()->getContentLanguage()->
            normalize( $stripped );
 
        return $normalized;
    }
 
    protected function startElement( $parser, string $name, array $attribs ) {
        $this->checkpointJustWritten = false;
 
        $this->clearOpenElement( null );
        $this->lastName = $name;
 
        if ( $name == 'revision' ) {
            $this->state = $name;
            $this->egress->writeOpenPage( null, $this->buffer );
            $this->buffer = "";
        } elseif ( $name == 'page' ) {
            $this->state = $name;
            if ( $this->atStart ) {
                $this->egress->writeOpenStream( $this->buffer );
                $this->buffer = "";
                $this->atStart = false;
            }
        } elseif ( $name === 'mediawiki' ) {
            if ( isset( $attribs['version'] ) ) {
                if ( $attribs['version'] !== $this->schemaVersion ) {
                    throw new RuntimeException( 'Mismatching schema version. '
                        . 'Use the --schema-version option to set the output schema version to '
                        . 'the version declared by the stub file, namely ' . $attribs['version'] );
                }
            }
        }
 
        if ( $name == "text" && ( isset( $attribs['id'] ) || isset( $attribs['location'] ) ) ) {
            $id = $attribs['location'] ?? $attribs['id'];
            $model = trim( $this->thisRevModel );
            $format = trim( $this->thisRevFormat );
 
            $model = $model === '' ? null : $model;
            $format = $format === '' ? null : $format;
            $expSize = !empty( $attribs['bytes'] ) && $model === CONTENT_MODEL_WIKITEXT
                ? (int)$attribs['bytes'] : null;
 
            $text = $this->getText( $id, $model, $format, $expSize );
 
            unset( $attribs['id'] );
            unset( $attribs['location'] );
            if ( $text !== '' ) {
                $attribs['xml:space'] = 'preserve';
            }
 
            $this->openElement = [ $name, $attribs ];
            if ( $text !== '' ) {
                $this->characterData( $parser, $text );
            }
        } else {
            $this->openElement = [ $name, $attribs ];
        }
    }
 
    protected function endElement( $parser, string $name ) {
        $this->checkpointJustWritten = false;
 
        if ( $this->openElement ) {
            $this->clearOpenElement( "" );
        } else {
            $this->buffer .= "</$name>";
        }
 
        if ( $name == 'revision' ) {
            $this->egress->writeRevision( null, $this->buffer );
            $this->buffer = "";
            $this->thisRev = "";
            $this->thisRole = null;
            $this->thisRevModel = null;
            $this->thisRevFormat = null;
        } elseif ( $name == 'page' ) {
            if ( !$this->firstPageWritten ) {
                $this->firstPageWritten = trim( $this->thisPage );
            }
            $this->lastPageWritten = trim( $this->thisPage );
            if ( $this->timeExceeded ) {
                $this->egress->writeClosePage( $this->buffer );
                // nasty hack, we can't just write the chardata after the
                // page tag, it will include leading blanks from the next line
                $this->egress->sink->write( "\n" );
 
                $this->buffer = $this->xmlwriterobj->closeStream();
                $this->egress->writeCloseStream( $this->buffer );
 
                $this->buffer = "";
                $this->thisPage = "";
                // this could be more than one file if we had more than one output arg
 
                $filenameList = (array)$this->egress->getFilenames();
                $newFilenames = [];
                $firstPageID = str_pad( $this->firstPageWritten, 9, "0", STR_PAD_LEFT );
                $lastPageID = str_pad( $this->lastPageWritten, 9, "0", STR_PAD_LEFT );
                $filenamesCount = count( $filenameList );
                for ( $i = 0; $i < $filenamesCount; $i++ ) {
                    $checkpointNameFilledIn = sprintf( $this->checkpointFiles[$i], $firstPageID, $lastPageID );
                    $fileinfo = pathinfo( $filenameList[$i] );
                    $newFilenames[] = $fileinfo['dirname'] . '/' . $checkpointNameFilledIn;
                }
                $this->egress->closeRenameAndReopen( $newFilenames );
                $this->buffer = $this->xmlwriterobj->openStream();
                $this->timeExceeded = false;
                $this->timeOfCheckpoint = $this->lastTime;
                $this->firstPageWritten = false;
                $this->checkpointJustWritten = true;
            } else {
                $this->egress->writeClosePage( $this->buffer );
                $this->buffer = "";
                $this->thisPage = "";
            }
        } elseif ( $name == 'mediawiki' ) {
            $this->egress->writeCloseStream( $this->buffer );
            $this->buffer = "";
        }
    }
 
    protected function characterData( $parser, string $data ) {
        $this->clearOpenElement( null );
        if ( $this->lastName == "id" ) {
            if ( $this->state == "revision" ) {
                $this->thisRev .= $data;
                $this->thisRole = SlotRecord::MAIN;
            } elseif ( $this->state == "page" ) {
                $this->thisPage .= $data;
            }
        } elseif ( $this->lastName == "model" ) {
            $this->thisRevModel .= $data;
        } elseif ( $this->lastName == "format" ) {
            $this->thisRevFormat .= $data;
        } elseif ( $this->lastName == "content" ) {
            $this->thisRole = "";
            $this->thisRevModel = "";
            $this->thisRevFormat = "";
        } elseif ( $this->lastName == "role" ) {
            $this->thisRole .= $data;
        }
 
        // have to skip the newline left over from closepagetag line of
        // end of checkpoint files. nasty hack!!
        if ( $this->checkpointJustWritten ) {
            if ( $data[0] == "\n" ) {
                $data = substr( $data, 1 );
            }
            $this->checkpointJustWritten = false;
        }
        $this->buffer .= htmlspecialchars( $data, ENT_COMPAT );
    }
 
    protected function clearOpenElement( ?string $style ) {
        if ( $this->openElement ) {
            $this->buffer .= Xml::element( $this->openElement[0], $this->openElement[1], $style );
            $this->openElement = false;
        }
    }
 
    private function isValidTextId( string $id ): bool {
        if ( preg_match( '/:/', $id ) ) {
            return $id !== 'tt:0';
        } elseif ( preg_match( '/^\d+$/', $id ) ) {
            return intval( $id ) > 0;
        }
 
        return false;
    }
 
}
 
class_alias( TextPassDumper::class, 'TextPassDumper' );