maintenance/backup.inc

   1 <?php
   2 /**
   3  * Base classes for database dumpers
   4  *
   5  * Copyright © 2005 Brion Vibber <brion@pobox.com>
   6  * https://www.mediawiki.org/
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  21  * http://www.gnu.org/copyleft/gpl.html
  22  *
  23  * @file
  24  * @ingroup Dump Maintenance
  25  */
  26
  27 require_once __DIR__ . '/Maintenance.php';
  28 require_once __DIR__ . '/../includes/export/DumpFilter.php';
  29
  30 use Wikimedia\Rdbms\LoadBalancer;
  31 use Wikimedia\Rdbms\IDatabase;
  32
  33 /**
  34  * @ingroup Dump Maintenance
  35  */
  36 class BackupDumper extends Maintenance {
  37         public $reporting = true;
  38         public $pages = null; // all pages
  39         public $skipHeader = false; // don't output <mediawiki> and <siteinfo>
  40         public $skipFooter = false; // don't output </mediawiki>
  41         public $startId = 0;
  42         public $endId = 0;
  43         public $revStartId = 0;
  44         public $revEndId = 0;
  45         public $dumpUploads = false;
  46         public $dumpUploadFileContents = false;
  47         public $orderRevs = false;
  48
  49         protected $reportingInterval = 100;
  50         protected $pageCount = 0;
  51         protected $revCount = 0;
  52         protected $server = null; // use default
  53         protected $sink = null; // Output filters
  54         protected $lastTime = 0;
  55         protected $pageCountLast = 0;
  56         protected $revCountLast = 0;
  57
  58         protected $outputTypes = [];
  59         protected $filterTypes = [];
  60
  61         protected $ID = 0;
  62
  63         /**
  64          * The dependency-injected database to use.
  65          *
  66          * @var IDatabase|null
  67          *
  68          * @see self::setDB
  69          */
  70         protected $forcedDb = null;
  71
  72         /** @var LoadBalancer */
  73         protected $lb;
  74
  75         // @todo Unused?
  76         private $stubText = false; // include rev_text_id instead of text; for 2-pass dump
  77
  78         /**
  79          * @param array $args For backward compatibility
  80          */
  81         function __construct( $args = null ) {
  82                 parent::__construct();
  83                 $this->stderr = fopen( "php://stderr", "wt" );
  84
  85                 // Built-in output and filter plugins
  86                 $this->registerOutput( 'file', 'DumpFileOutput' );
  87                 $this->registerOutput( 'gzip', 'DumpGZipOutput' );
  88                 $this->registerOutput( 'bzip2', 'DumpBZip2Output' );
  89                 $this->registerOutput( 'dbzip2', 'DumpDBZip2Output' );
  90                 $this->registerOutput( '7zip', 'Dump7ZipOutput' );
  91
  92                 $this->registerFilter( 'latest', 'DumpLatestFilter' );
  93                 $this->registerFilter( 'notalk', 'DumpNotalkFilter' );
  94                 $this->registerFilter( 'namespace', 'DumpNamespaceFilter' );
  95
  96                 // These three can be specified multiple times
  97                 $this->addOption( 'plugin', 'Load a dump plugin class. Specify as <class>[:<file>].',
  98                         false, true, false, true );
  99                 $this->addOption( 'output', 'Begin a filtered output stream; Specify as <type>:<file>. ' .
 100                         '<type>s: file, gzip, bzip2, 7zip, dbzip2', false, true, false, true );
 101                 $this->addOption( 'filter', 'Add a filter on an output branch. Specify as ' .
 102                         '<type>[:<options>]. <types>s: latest, notalk, namespace', false, true, false, true );
 103                 $this->addOption( 'report', 'Report position and speed after every n pages processed. ' .
 104                         'Default: 100.', false, true );
 105                 $this->addOption( 'server', 'Force reading from MySQL server', false, true );
 106                 $this->addOption( '7ziplevel', '7zip compression level for all 7zip outputs. Used for ' .
 107                         '-mx option to 7za command.', false, true );
 108
 109                 if ( $args ) {
 110                         // Args should be loaded and processed so that dump() can be called directly
 111                         // instead of execute()
 112                         $this->loadWithArgv( $args );
 113                         $this->processOptions();
 114                 }
 115         }
 116
 117         /**
 118          * @param string $name
 119          * @param string $class Name of output filter plugin class
 120          */
 121         function registerOutput( $name, $class ) {
 122                 $this->outputTypes[$name] = $class;
 123         }
 124
 125         /**
 126          * @param string $name
 127          * @param string $class Name of filter plugin class
 128          */
 129         function registerFilter( $name, $class ) {
 130                 $this->filterTypes[$name] = $class;
 131         }
 132
 133         /**
 134          * Load a plugin and register it
 135          *
 136          * @param string $class Name of plugin class; must have a static 'register'
 137          *   method that takes a BackupDumper as a parameter.
 138          * @param string $file Full or relative path to the PHP file to load, or empty
 139          */
 140         function loadPlugin( $class, $file ) {
 141                 if ( $file != '' ) {
 142                         require_once $file;
 143                 }
 144                 $register = [ $class, 'register' ];
 145                 call_user_func_array( $register, [ $this ] );
 146         }
 147
 148         function execute() {
 149                 throw new MWException( 'execute() must be overridden in subclasses' );
 150         }
 151
 152         /**
 153          * Processes arguments and sets $this->$sink accordingly
 154          */
 155         function processOptions() {
 156                 $sink = null;
 157                 $sinks = [];
 158
 159                 $options = $this->orderedOptions;
 160                 foreach ( $options as $arg ) {
 161                         $opt = $arg[0];
 162                         $param = $arg[1];
 163
 164                         switch ( $opt ) {
 165                                 case 'plugin':
 166                                         $val = explode( ':', $param );
 167
 168                                         if ( count( $val ) === 1 ) {
 169                                                 $this->loadPlugin( $val[0], '' );
 170                                         } elseif ( count( $val ) === 2 ) {
 171                                                 $this->loadPlugin( $val[0], $val[1] );
 172                                         } else {
 173                                                 $this->fatalError( 'Invalid plugin parameter' );
 174                                                 return;
 175                                         }
 176
 177                                         break;
 178                                 case 'output':
 179                                         $split = explode( ':', $param, 2 );
 180                                         if ( count( $split ) !== 2 ) {
 181                                                 $this->fatalError( 'Invalid output parameter' );
 182                                         }
 183                                         list( $type, $file ) = $split;
 184                                         if ( !is_null( $sink ) ) {
 185                                                 $sinks[] = $sink;
 186                                         }
 187                                         if ( !isset( $this->outputTypes[$type] ) ) {
 188                                                 $this->fatalError( "Unrecognized output sink type '$type'" );
 189                                         }
 190                                         $class = $this->outputTypes[$type];
 191                                         if ( $type === "7zip" ) {
 192                                                 $sink = new $class( $file, intval( $this->getOption( '7ziplevel' ) ) );
 193                                         } else {
 194                                                 $sink = new $class( $file );
 195                                         }
 196
 197                                         break;
 198                                 case 'filter':
 199                                         if ( is_null( $sink ) ) {
 200                                                 $sink = new DumpOutput();
 201                                         }
 202
 203                                         $split = explode( ':', $param );
 204                                         $key = $split[0];
 205
 206                                         if ( !isset( $this->filterTypes[$key] ) ) {
 207                                                 $this->fatalError( "Unrecognized filter type '$key'" );
 208                                         }
 209
 210                                         $type = $this->filterTypes[$key];
 211
 212                                         if ( count( $split ) === 1 ) {
 213                                                 $filter = new $type( $sink );
 214                                         } elseif ( count( $split ) === 2 ) {
 215                                                 $filter = new $type( $sink, $split[1] );
 216                                         } else {
 217                                                 $this->fatalError( 'Invalid filter parameter' );
 218                                         }
 219
 220                                         // references are lame in php...
 221                                         unset( $sink );
 222                                         $sink = $filter;
 223
 224                                         break;
 225                         }
 226                 }
 227
 228                 if ( $this->hasOption( 'report' ) ) {
 229                         $this->reportingInterval = intval( $this->getOption( 'report' ) );
 230                 }
 231
 232                 if ( $this->hasOption( 'server' ) ) {
 233                         $this->server = $this->getOption( 'server' );
 234                 }
 235
 236                 if ( is_null( $sink ) ) {
 237                         $sink = new DumpOutput();
 238                 }
 239                 $sinks[] = $sink;
 240
 241                 if ( count( $sinks ) > 1 ) {
 242                         $this->sink = new DumpMultiWriter( $sinks );
 243                 } else {
 244                         $this->sink = $sink;
 245                 }
 246         }
 247
 248         function dump( $history, $text = WikiExporter::TEXT ) {
 249                 # Notice messages will foul up your XML output even if they're
 250                 # relatively harmless.
 251                 if ( ini_get( 'display_errors' ) ) {
 252                         ini_set( 'display_errors', 'stderr' );
 253                 }
 254
 255                 $this->initProgress( $history );
 256
 257                 $db = $this->backupDb();
 258                 $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
 259                 $exporter->dumpUploads = $this->dumpUploads;
 260                 $exporter->dumpUploadFileContents = $this->dumpUploadFileContents;
 261
 262                 $wrapper = new ExportProgressFilter( $this->sink, $this );
 263                 $exporter->setOutputSink( $wrapper );
 264
 265                 if ( !$this->skipHeader ) {
 266                         $exporter->openStream();
 267                 }
 268                 # Log item dumps: all or by range
 269                 if ( $history & WikiExporter::LOGS ) {
 270                         if ( $this->startId || $this->endId ) {
 271                                 $exporter->logsByRange( $this->startId, $this->endId );
 272                         } else {
 273                                 $exporter->allLogs();
 274                         }
 275                 } elseif ( is_null( $this->pages ) ) {
 276                         # Page dumps: all or by page ID range
 277                         if ( $this->startId || $this->endId ) {
 278                                 $exporter->pagesByRange( $this->startId, $this->endId, $this->orderRevs );
 279                         } elseif ( $this->revStartId || $this->revEndId ) {
 280                                 $exporter->revsByRange( $this->revStartId, $this->revEndId );
 281                         } else {
 282                                 $exporter->allPages();
 283                         }
 284                 } else {
 285                         # Dump of specific pages
 286                         $exporter->pagesByName( $this->pages );
 287                 }
 288
 289                 if ( !$this->skipFooter ) {
 290                         $exporter->closeStream();
 291                 }
 292
 293                 $this->report( true );
 294         }
 295
 296         /**
 297          * Initialise starting time and maximum revision count.
 298          * We'll make ETA calculations based an progress, assuming relatively
 299          * constant per-revision rate.
 300          * @param int $history WikiExporter::CURRENT or WikiExporter::FULL
 301          */
 302         function initProgress( $history = WikiExporter::FULL ) {
 303                 $table = ( $history == WikiExporter::CURRENT ) ? 'page' : 'revision';
 304                 $field = ( $history == WikiExporter::CURRENT ) ? 'page_id' : 'rev_id';
 305
 306                 $dbr = $this->forcedDb;
 307                 if ( $this->forcedDb === null ) {
 308                         $dbr = wfGetDB( DB_REPLICA );
 309                 }
 310                 $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', __METHOD__ );
 311                 $this->startTime = microtime( true );
 312                 $this->lastTime = $this->startTime;
 313                 $this->ID = getmypid();
 314         }
 315
 316         /**
 317          * @todo Fixme: the --server parameter is currently not respected, as it
 318          * doesn't seem terribly easy to ask the load balancer for a particular
 319          * connection by name.
 320          * @return IDatabase
 321          */
 322         function backupDb() {
 323                 if ( $this->forcedDb !== null ) {
 324                         return $this->forcedDb;
 325                 }
 326
 327                 $this->lb = wfGetLBFactory()->newMainLB();
 328                 $db = $this->lb->getConnection( DB_REPLICA, 'dump' );
 329
 330                 // Discourage the server from disconnecting us if it takes a long time
 331                 // to read out the big ol' batch query.
 332                 $db->setSessionOptions( [ 'connTimeout' => 3600 * 24 ] );
 333
 334                 return $db;
 335         }
 336
 337         /**
 338          * Force the dump to use the provided database connection for database
 339          * operations, wherever possible.
 340          *
 341          * @param IDatabase|null $db (Optional) the database connection to use. If null, resort to
 342          *   use the globally provided ways to get database connections.
 343          */
 344         function setDB( IDatabase $db = null ) {
 345                 parent::setDB( $db );
 346                 $this->forcedDb = $db;
 347         }
 348
 349         function __destruct() {
 350                 if ( isset( $this->lb ) ) {
 351                         $this->lb->closeAll();
 352                 }
 353         }
 354
 355         function backupServer() {
 356                 global $wgDBserver;
 357
 358                 return $this->server
 359                         ? $this->server
 360                         : $wgDBserver;
 361         }
 362
 363         function reportPage() {
 364                 $this->pageCount++;
 365         }
 366
 367         function revCount() {
 368                 $this->revCount++;
 369                 $this->report();
 370         }
 371
 372         function report( $final = false ) {
 373                 if ( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
 374                         $this->showReport();
 375                 }
 376         }
 377
 378         function showReport() {
 379                 if ( $this->reporting ) {
 380                         $now = wfTimestamp( TS_DB );
 381                         $nowts = microtime( true );
 382                         $deltaAll = $nowts - $this->startTime;
 383                         $deltaPart = $nowts - $this->lastTime;
 384                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 385                         $this->revCountPart = $this->revCount - $this->revCountLast;
 386
 387                         if ( $deltaAll ) {
 388                                 $portion = $this->revCount / $this->maxCount;
 389                                 $eta = $this->startTime + $deltaAll / $portion;
 390                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 391                                 $pageRate = $this->pageCount / $deltaAll;
 392                                 $revRate = $this->revCount / $deltaAll;
 393                         } else {
 394                                 $pageRate = '-';
 395                                 $revRate = '-';
 396                                 $etats = '-';
 397                         }
 398                         if ( $deltaPart ) {
 399                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 400                                 $revRatePart = $this->revCountPart / $deltaPart;
 401                         } else {
 402                                 $pageRatePart = '-';
 403                                 $revRatePart = '-';
 404                         }
 405                         $this->progress( sprintf(
 406                                 "%s: %s (ID %d) %d pages (%0.1f|%0.1f/sec all|curr), "
 407                                         . "%d revs (%0.1f|%0.1f/sec all|curr), ETA %s [max %d]",
 408                                 $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate,
 409                                 $pageRatePart, $this->revCount, $revRate, $revRatePart, $etats,
 410                                 $this->maxCount
 411                         ) );
 412                         $this->lastTime = $nowts;
 413                         $this->revCountLast = $this->revCount;
 414                 }
 415         }
 416
 417         function progress( $string ) {
 418                 if ( $this->reporting ) {
 419                         fwrite( $this->stderr, $string . "\n" );
 420                 }
 421         }
 422
 423         function fatalError( $msg ) {
 424                 $this->error( "$msg\n", 1 );
 425         }
 426 }
 427
 428 class ExportProgressFilter extends DumpFilter {
 429         function __construct( &$sink, &$progress ) {
 430                 parent::__construct( $sink );
 431                 $this->progress = $progress;
 432         }
 433
 434         function writeClosePage( $string ) {
 435                 parent::writeClosePage( $string );
 436                 $this->progress->reportPage();
 437         }
 438
 439         function writeRevision( $rev, $string ) {
 440                 parent::writeRevision( $rev, $string );
 441                 $this->progress->revCount();
 442         }
 443 }