/**
* Script that postprocesses XML dumps from dumpBackup.php to add page text
*
- * Copyright © 2005 Brion Vibber <brion@pobox.com>, 2010 Alexandre Emsenhuber
+ * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
* http://www.mediawiki.org/
*
* This program is free software; you can redistribute it and/or modify
$originalDir = getcwd();
-require_once( dirname( __FILE__ ) . '/commandLine.inc' );
-require_once( 'backup.inc' );
-
-/**
- * @ingroup Maintenance
- */
-class TextPassDumper extends BackupDumper {
- var $prefetch = null;
- var $input = "php://stdin";
- var $fetchCount = 0;
- var $prefetchCount = 0;
- var $lastTime = 0;
- var $pageCountLast = 0;
- var $revCountLast = 0;
- var $prefetchCountLast = 0;
- var $fetchCountLast = 0;
-
- var $failures = 0;
- var $maxFailures = 5;
- var $failedTextRetrievals = 0;
- var $maxConsecutiveFailedTextRetrievals = 200;
- var $failureTimeout = 5; // Seconds to sleep after db failure
-
- var $php = "php";
- var $spawn = false;
- var $spawnProc = false;
- var $spawnWrite = false;
- var $spawnRead = false;
- var $spawnErr = false;
-
- var $ID = 0;
-
- function initProgress( $history ) {
- parent::initProgress();
- $this->ID = getmypid();
- $this->lastTime = $this->startTime;
- }
-
- function dump( $history, $text = WikiExporter::TEXT ) {
- # This shouldn't happen if on console... ;)
- header( 'Content-type: text/html; charset=UTF-8' );
-
- # 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->db = $this->backupDb();
-
- $this->readDump();
-
- if ( $this->spawnProc ) {
- $this->closeSpawn();
- }
-
- $this->report( true );
- }
-
- function processOption( $opt, $val, $param ) {
- global $IP;
- $url = $this->processFileOpt( $val, $param );
-
- switch( $opt ) {
- case 'prefetch':
- require_once "$IP/maintenance/backupPrefetch.inc";
- $this->prefetch = new BaseDump( $url );
- break;
- case 'stub':
- $this->input = $url;
- break;
- case 'spawn':
- $this->spawn = true;
- if ( $val ) {
- $this->php = $val;
- }
- break;
- }
- }
-
- function processFileOpt( $val, $param ) {
- $fileURIs = explode(';',$param);
- foreach ( $fileURIs as $URI ) {
- switch( $val ) {
- case "file":
- $newURI = $URI;
- break;
- case "gzip":
- $newURI = "compress.zlib://$URI";
- break;
- case "bzip2":
- $newURI = "compress.bzip2://$URI";
- break;
- case "7zip":
- $newURI = "mediawiki.compress.7z://$URI";
- break;
- default:
- $newURI = $URI;
- }
- $newFileURIs[] = $newURI;
- }
- $val = implode( ';', $newFileURIs );
- return $val;
- }
-
- /**
- * Overridden to include prefetch ratio if enabled.
- */
- function showReport() {
- if ( !$this->prefetch ) {
- return parent::showReport();
- }
-
- if ( $this->reporting ) {
- $now = wfTimestamp( TS_DB );
- $deltaAll = wfTime() - $this->startTime;
- $deltaPart = wfTime() - $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 = '-';
- }
- $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, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $fetchRate, $fetchRatePart, $etats, $this->maxCount ) );
- $this->lastTime = $now;
- $this->partCountLast = $this->partCount;
- $this->revCountLast = $this->revCount;
- $this->prefetchCountLast = $this->prefetchCount;
- $this->fetchCountLast = $this->fetchCount;
- }
- }
-
- function readDump() {
- $state = '';
- $lastName = '';
- $this->thisPage = 0;
- $this->thisRev = 0;
-
- $reader = new XMLReader();
- $reader->open( $this->input );
- $writer = new XMLWriter();
- $writer->openMemory();
-
-
- while ( $reader->read() ) {
- $tag = $reader->name;
- $type = $reader->nodeType;
-
- if ( $type == XmlReader::END_ELEMENT ) {
- $writer->endElement();
-
- if ( $tag == 'revision' ) {
- $this->revCount();
- $this->thisRev = '';
- } elseif ( $tag == 'page' ) {
- $this->reportPage();
- $this->thisPage = '';
- }
- } elseif ( $type == XmlReader::ELEMENT ) {
- $attribs = array();
- if ( $reader->hasAttributes ) {
- for ( $i = 0; $reader->moveToAttributeNo( $i ); $i++ ) {
- $attribs[$reader->name] = $reader->value;
- }
- }
-
- if ( $reader->isEmptyElement && $tag == 'text' && isset( $attribs['id'] ) ) {
- $writer->startElement( 'text' );
- $writer->writeAttribute( 'xml:space', 'preserve' );
- $text = $this->getText( $attribs['id'] );
- if ( strlen( $text ) ) {
- $writer->text( $text );
- }
- $writer->endElement();
- } else {
- $writer->startElement( $tag );
- foreach( $attribs as $name => $val ) {
- $writer->writeAttribute( $name, $val );
- }
- if ( $reader->isEmptyElement ) {
- $writer->endElement();
- }
- }
-
- $lastName = $tag;
- if ( $tag == 'revision' ) {
- $state = 'revision';
- } elseif ( $tag == 'page' ) {
- $state = 'page';
- }
- } elseif ( $type == XMLReader::SIGNIFICANT_WHITESPACE || $type = XMLReader::TEXT ) {
- if ( $lastName == 'id' ) {
- if ( $state == 'revision' ) {
- $this->thisRev .= $reader->value;
- } elseif ( $state == 'page' ) {
- $this->thisPage .= $reader->value;
- }
- }
- $writer->text( $reader->value );
- }
- $this->sink->write( $writer->outputMemory() );
- }
- }
-
- function getText( $id ) {
- $this->fetchCount++;
- if ( isset( $this->prefetch ) ) {
- $text = $this->prefetch->prefetch( $this->thisPage, $this->thisRev );
- if ( $text !== null ) { // Entry missing from prefetch dump
- $dbr = wfGetDB( DB_SLAVE );
- $revID = intval( $this->thisRev );
- $revLength = $dbr->selectField( 'revision', 'rev_len', array( 'rev_id' => $revID ) );
- // if length of rev text in file doesn't match length in db, we reload
- // this avoids carrying forward broken data from previous xml dumps
- if( strlen( $text ) == $revLength ) {
- $this->prefetchCount++;
- return $text;
- }
- }
- }
- return $this->doGetText( $id );
- }
-
- private function doGetText( $id ) {
- $id = intval( $id );
- $this->failures = 0;
- $ex = new MWException( "Graceful storage failure" );
- while (true) {
- if ( $this->spawn ) {
- if ($this->failures) {
- // we don't know why it failed, could be the child process
- // borked, could be db entry busted, could be db server out to lunch,
- // so cover all bases
- $this->closeSpawn();
- $this->openSpawn();
- }
- $text = $this->getTextSpawned( $id );
- } else {
- $text = $this->getTextDbSafe( $id );
- }
- if ( $text === false ) {
- $this->failures++;
- if ( $this->failures > $this->maxFailures) {
- $this->progress( "Failed to retrieve revision text for text id ".
- "$id after $this->maxFailures tries, giving up" );
- // were there so many bad retrievals in a row we want to bail?
- // at some point we have to declare the dump irretrievably broken
- $this->failedTextRetrievals++;
- if ($this->failedTextRetrievals > $this->maxConsecutiveFailedTextRetrievals) {
- throw $ex;
- }
- else {
- // would be nice to return something better to the caller someday,
- // log what we know about the failure and about the revision
- return("");
- }
- } else {
- $this->progress( "Error $this->failures " .
- "of allowed $this->maxFailures retrieving revision text for text id $id! " .
- "Pausing $this->failureTimeout seconds before retry..." );
- sleep( $this->failureTimeout );
- }
- } else {
- $this->failedTextRetrievals= 0;
- return( $text );
- }
- }
-
- }
-
- /**
- * Fetch a text revision from the database, retrying in case of failure.
- * This may survive some transitory errors by reconnecting, but
- * may not survive a long-term server outage.
- */
- private function getTextDbSafe( $id ) {
- while ( true ) {
- try {
- $text = $this->getTextDb( $id );
- } catch ( DBQueryError $ex ) {
- $text = false;
- }
- return $text;
- }
- }
-
- /**
- * May throw a database error if, say, the server dies during query.
- */
- private function getTextDb( $id ) {
- global $wgContLang;
- $row = $this->db->selectRow( 'text',
- array( 'old_text', 'old_flags' ),
- array( 'old_id' => $id ),
- __METHOD__ );
- $text = Revision::getRevisionText( $row );
- if ( $text === false ) {
- return false;
- }
- $stripped = str_replace( "\r", "", $text );
- $normalized = $wgContLang->normalize( $stripped );
- return $normalized;
- }
-
- private function getTextSpawned( $id ) {
- wfSuppressWarnings();
- if ( !$this->spawnProc ) {
- // First time?
- $this->openSpawn();
- }
- $text = $this->getTextSpawnedOnce( $id );
- wfRestoreWarnings();
- return $text;
- }
-
- function openSpawn() {
- global $IP;
-
- $cmd = implode( " ",
- array_map( 'wfEscapeShellArg',
- array(
- $this->php,
- "$IP/maintenance/fetchText.php",
- '--wiki', wfWikiID() ) ) );
- $spec = array(
- 0 => array( "pipe", "r" ),
- 1 => array( "pipe", "w" ),
- 2 => array( "file", "/dev/null", "a" ) );
- $pipes = array();
-
- $this->progress( "Spawning database subprocess: $cmd" );
- $this->spawnProc = proc_open( $cmd, $spec, $pipes );
- if ( !$this->spawnProc ) {
- // shit
- $this->progress( "Subprocess spawn failed." );
- return false;
- }
- list(
- $this->spawnWrite, // -> stdin
- $this->spawnRead, // <- stdout
- ) = $pipes;
-
- return true;
- }
-
- private function closeSpawn() {
- wfSuppressWarnings();
- if ( $this->spawnRead )
- fclose( $this->spawnRead );
- $this->spawnRead = false;
- if ( $this->spawnWrite )
- fclose( $this->spawnWrite );
- $this->spawnWrite = false;
- if ( $this->spawnErr )
- fclose( $this->spawnErr );
- $this->spawnErr = false;
- if ( $this->spawnProc )
- pclose( $this->spawnProc );
- $this->spawnProc = false;
- wfRestoreWarnings();
- }
-
- private function getTextSpawnedOnce( $id ) {
- global $wgContLang;
-
- $ok = fwrite( $this->spawnWrite, "$id\n" );
- // $this->progress( ">> $id" );
- if ( !$ok ) return false;
-
- $ok = fflush( $this->spawnWrite );
- // $this->progress( ">> [flush]" );
- if ( !$ok ) return false;
-
- // check that the text id they are sending is the one we asked for
- // this avoids out of sync revision text errors we have encountered in the past
- $newId = fgets( $this->spawnRead );
- if ( $newId === false ) {
- return false;
- }
- if ( $id != intval( $newId ) ) {
- 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 = $wgContLang->normalize( $stripped );
- return $normalized;
- }
-}
+require_once( __DIR__ . '/commandLine.inc' );
+require_once( __DIR__ . '/backupTextPass.inc' );
$dumper = new TextPassDumper( $argv );
if ( !isset( $options['help'] ) ) {
- $dumper->dump( WikiExporter::FULL );
+ $dumper->dump( true );
} else {
$dumper->progress( <<<ENDS
This script postprocesses XML dumps from dumpBackup.php to add
--stub=<type>:<file> To load a compressed stub dump instead of stdin
--prefetch=<type>:<file> Use a prior dump file as a text source, to save
pressure on the database.
+ (Requires the XMLReader extension)
+ --maxtime=<minutes> 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.
+ --checkpointfile=<filenamepattern> 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.
--quiet Don't dump status reports to stderr.
--report=n Report position and speed after every n pages processed.
(Default: 100)
--server=h Force reading from MySQL server h
- --output=<type>:<file> Write to a file instead of stdout
- <type>s: file, gzip, bzip2, 7zip
--current Base ETA on number of pages in database instead of all revisions
--spawn Spawn a subprocess for loading text records
--help Display this help message
ENDS
);
}
-
-