<?php
/**
- * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
+ * Base classes for database dumpers
+ *
+ * Copyright © 2005 Brion Vibber <brion@pobox.com>
* http://www.mediawiki.org/
*
* This program is free software; you can redistribute it and/or modify
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
* http://www.gnu.org/copyleft/gpl.html
*
- * @package MediaWiki
- * @subpackage SpecialPage
+ * @file
+ * @ingroup Dump Maintenance
*/
+/**
+ * @ingroup Dump Maintenance
+ */
class DumpDBZip2Output extends DumpPipeOutput {
- function DumpDBZip2Output( $file ) {
- parent::DumpPipeOutput( "dbzip2", $file );
+ function __construct( $file ) {
+ parent::__construct( "dbzip2", $file );
}
}
+/**
+ * @ingroup Dump Maintenance
+ */
class BackupDumper {
var $reportingInterval = 100;
var $reporting = true;
var $skipFooter = false; // don't output </mediawiki>
var $startId = 0;
var $endId = 0;
+ var $revStartId = 0;
+ var $revEndId = 0;
var $sink = null; // Output filters
var $stubText = false; // include rev_text_id instead of text; for 2-pass dump
+ var $dumpUploads = false;
+ var $dumpUploadFileContents = false;
+ var $lastTime = 0;
+ var $pageCountLast = 0;
+ var $revCountLast = 0;
+ var $ID = 0;
- function BackupDumper( $args ) {
+ var $outputTypes = array(), $filterTypes = array();
+
+ /**
+ * The dependency-injected database to use.
+ *
+ * @var DatabaseBase|null
+ *
+ * @see self::setDb
+ */
+ protected $forcedDb = null;
+
+ /**
+ * @var LoadBalancer
+ */
+ protected $lb;
+
+ function __construct( $args ) {
$this->stderr = fopen( "php://stderr", "wt" );
// Built-in output and filter plugins
}
/**
- * @param string $name
- * @param string $class name of output filter plugin class
+ * @param $name String
+ * @param $class String: name of output filter plugin class
*/
function registerOutput( $name, $class ) {
$this->outputTypes[$name] = $class;
}
/**
- * @param string $name
- * @param string $class name of filter plugin class
+ * @param $name String
+ * @param $class String: name of filter plugin class
*/
function registerFilter( $name, $class ) {
$this->filterTypes[$name] = $class;
/**
* Load a plugin and register it
- * @param string $class Name of plugin class; must have a static 'register'
- * method that takes a BackupDumper as a parameter.
- * @param string $file Full or relative path to the PHP file to load, or empty
+ *
+ * @param $class String: name of plugin class; must have a static 'register'
+ * method that takes a BackupDumper as a parameter.
+ * @param $file String: full or relative path to the PHP file to load, or empty
*/
function loadPlugin( $class, $file ) {
- if( $file != '' ) {
+ if ( $file != '' ) {
require_once( $file );
}
$register = array( $class, 'register' );
}
/**
- * @param array $args
- * @return array
- * @static
+ * @param $args Array
+ * @return Array
*/
function processArgs( $args ) {
$sink = null;
$sinks = array();
- foreach( $args as $arg ) {
- if( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
- @list( $full, $opt, $val, $param ) = $matches;
+ foreach ( $args as $arg ) {
+ $matches = array();
+ if ( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
+ @list( /* $full */ , $opt, $val, $param ) = $matches;
switch( $opt ) {
case "plugin":
$this->loadPlugin( $val, $param );
break;
case "output":
- if( !is_null( $sink ) ) {
+ if ( !is_null( $sink ) ) {
$sinks[] = $sink;
}
- if( !isset( $this->outputTypes[$val] ) ) {
- wfDie( "Unrecognized output sink type '$val'\n" );
+ if ( !isset( $this->outputTypes[$val] ) ) {
+ $this->fatalError( "Unrecognized output sink type '$val'" );
}
$type = $this->outputTypes[$val];
$sink = new $type( $param );
break;
case "filter":
- if( is_null( $sink ) ) {
- $this->progress( "Warning: assuming stdout for filter output\n" );
+ if ( is_null( $sink ) ) {
$sink = new DumpOutput();
}
- if( !isset( $this->filterTypes[$val] ) ) {
- wfDie( "Unrecognized filter type '$val'\n" );
+ if ( !isset( $this->filterTypes[$val] ) ) {
+ $this->fatalError( "Unrecognized filter type '$val'" );
}
$type = $this->filterTypes[$val];
$filter = new $type( $sink, $param );
$this->server = $val;
break;
case "force-normal":
- if( !function_exists( 'utf8_normalize' ) ) {
- dl( "php_utfnormal.so" );
- if( !function_exists( 'utf8_normalize' ) ) {
- wfDie( "Failed to load UTF-8 normalization extension. " .
- "Install or remove --force-normal parameter to use slower code.\n" );
+ if ( !function_exists( 'utf8_normalize' ) ) {
+ wfDl( "php_utfnormal.so" );
+ if ( !function_exists( 'utf8_normalize' ) ) {
+ $this->fatalError( "Failed to load UTF-8 normalization extension. " .
+ "Install or remove --force-normal parameter to use slower code." );
}
}
break;
}
}
- if( is_null( $sink ) ) {
+ if ( is_null( $sink ) ) {
$sink = new DumpOutput();
}
$sinks[] = $sink;
- if( count( $sinks ) > 1 ) {
+ if ( count( $sinks ) > 1 ) {
return new DumpMultiWriter( $sinks );
} else {
return $sink;
// extension point for subclasses to add options
}
- function dump( $history, $text = MW_EXPORT_TEXT ) {
+ function dump( $history, $text = WikiExporter::TEXT ) {
# Notice messages will foul up your XML output even if they're
# relatively harmless.
- ini_set( 'display_errors', false );
+ if ( ini_get( 'display_errors' ) )
+ ini_set( 'display_errors', 'stderr' );
$this->initProgress( $history );
- $db =& $this->backupDb();
- $exporter = new WikiExporter( $db, $history, MW_EXPORT_STREAM, $text );
+ $db = $this->backupDb();
+ $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
+ $exporter->dumpUploads = $this->dumpUploads;
+ $exporter->dumpUploadFileContents = $this->dumpUploadFileContents;
$wrapper = new ExportProgressFilter( $this->sink, $this );
$exporter->setOutputSink( $wrapper );
- if( !$this->skipHeader )
+ if ( !$this->skipHeader )
$exporter->openStream();
-
- if( is_null( $this->pages ) ) {
- if( $this->startId || $this->endId ) {
+ # Log item dumps: all or by range
+ if ( $history & WikiExporter::LOGS ) {
+ if ( $this->startId || $this->endId ) {
+ $exporter->logsByRange( $this->startId, $this->endId );
+ } else {
+ $exporter->allLogs();
+ }
+ # Page dumps: all or by page ID range
+ } else if ( is_null( $this->pages ) ) {
+ if ( $this->startId || $this->endId ) {
$exporter->pagesByRange( $this->startId, $this->endId );
+ } elseif ( $this->revStartId || $this->revEndId ) {
+ $exporter->revsByRange( $this->revStartId, $this->revEndId );
} else {
$exporter->allPages();
}
+ # Dump of specific pages
} else {
$exporter->pagesByName( $this->pages );
}
- if( !$this->skipFooter )
+ if ( !$this->skipFooter )
$exporter->closeStream();
$this->report( true );
}
-
+
/**
* Initialise starting time and maximum revision count.
* We'll make ETA calculations based an progress, assuming relatively
* constant per-revision rate.
- * @param int $history MW_EXPORT_CURRENT or MW_EXPORT_FULL
+ * @param $history Integer: WikiExporter::CURRENT or WikiExporter::FULL
*/
- function initProgress( $history = MW_EXPORT_FULL ) {
- $table = ($history == MW_EXPORT_CURRENT) ? 'page' : 'revision';
- $field = ($history == MW_EXPORT_CURRENT) ? 'page_id' : 'rev_id';
-
- $dbr =& wfGetDB( DB_SLAVE );
- $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', 'BackupDumper::dump' );
- $this->startTime = wfTime();
+ function initProgress( $history = WikiExporter::FULL ) {
+ $table = ( $history == WikiExporter::CURRENT ) ? 'page' : 'revision';
+ $field = ( $history == WikiExporter::CURRENT ) ? 'page_id' : 'rev_id';
+
+ $dbr = $this->forcedDb;
+ if ( $this->forcedDb === null ) {
+ $dbr = wfGetDB( DB_SLAVE );
+ }
+ $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', __METHOD__ );
+ $this->startTime = microtime( true );
+ $this->lastTime = $this->startTime;
+ $this->ID = getmypid();
}
- function &backupDb() {
- global $wgDBadminuser, $wgDBadminpassword;
- global $wgDBname, $wgDebugDumpSql;
- $flags = ($wgDebugDumpSql ? DBO_DEBUG : 0) | DBO_DEFAULT; // god-damn hack
- $db = new Database( $this->backupServer(), $wgDBadminuser, $wgDBadminpassword, $wgDBname, false, $flags );
- $timeout = 3600 * 24;
- $db->query( "SET net_read_timeout=$timeout" );
- $db->query( "SET net_write_timeout=$timeout" );
+ /**
+ * @todo Fixme: the --server parameter is currently not respected, as it
+ * doesn't seem terribly easy to ask the load balancer for a particular
+ * connection by name.
+ * @return DatabaseBase
+ */
+ function backupDb() {
+ if ( $this->forcedDb !== null ) {
+ return $this->forcedDb;
+ }
+
+ $this->lb = wfGetLBFactory()->newMainLB();
+ $db = $this->lb->getConnection( DB_SLAVE, 'backup' );
+
+ // Discourage the server from disconnecting us if it takes a long time
+ // to read out the big ol' batch query.
+ $db->setSessionOptions( array( 'connTimeout' => 3600 * 24 ) );
+
return $db;
}
+ /**
+ * Force the dump to use the provided database connection for database
+ * operations, wherever possible.
+ *
+ * @param $db DatabaseBase|null: (Optional) the database connection to
+ * use. If null, resort to use the globally provided ways to
+ * get database connections.
+ */
+ function setDb( DatabaseBase $db = null ) {
+ $this->forcedDb = $db;
+ }
+
+ function __destruct() {
+ if ( isset( $this->lb ) ) {
+ $this->lb->closeAll();
+ }
+ }
+
function backupServer() {
global $wgDBserver;
return $this->server
}
function report( $final = false ) {
- if( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
+ if ( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
$this->showReport();
}
}
function showReport() {
- if( $this->reporting ) {
- $delta = wfTime() - $this->startTime;
+ if ( $this->reporting ) {
$now = wfTimestamp( TS_DB );
- if( $delta ) {
- $rate = $this->pageCount / $delta;
- $revrate = $this->revCount / $delta;
+ $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 + $delta / $portion;
+ $eta = $this->startTime + $deltaAll / $portion;
$etats = wfTimestamp( TS_DB, intval( $eta ) );
+ $pageRate = $this->pageCount / $deltaAll;
+ $revRate = $this->revCount / $deltaAll;
} else {
- $rate = '-';
- $revrate = '-';
+ $pageRate = '-';
+ $revRate = '-';
$etats = '-';
}
- global $wgDBname;
- $this->progress( sprintf( "%s: %s %d pages (%0.3f/sec), %d revs (%0.3f/sec), ETA %s [max %d]",
- $now, $wgDBname, $this->pageCount, $rate, $this->revCount, $revrate, $etats, $this->maxCount ) );
+ if ( $deltaPart ) {
+ $pageRatePart = $this->pageCountPart / $deltaPart;
+ $revRatePart = $this->revCountPart / $deltaPart;
+ } else {
+ $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), ETA %s [max %d]",
+ $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $etats, $this->maxCount ) );
+ $this->lastTime = $nowts;
+ $this->revCountLast = $this->revCount;
}
}
function progress( $string ) {
fwrite( $this->stderr, $string . "\n" );
}
+
+ function fatalError( $msg ) {
+ $this->progress( "$msg\n" );
+ die(1);
+ }
}
class ExportProgressFilter extends DumpFilter {
- function ExportProgressFilter( &$sink, &$progress ) {
- parent::DumpFilter( $sink );
+ function __construct( &$sink, &$progress ) {
+ parent::__construct( $sink );
$this->progress = $progress;
}
$this->progress->revCount();
}
}
-
-?>