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