maintenance/backup.inc

   1 <?php
   2 /**
   3  * Base classes for database dumpers
   4  *
   5  * Copyright © 2005 Brion Vibber <brion@pobox.com>
   6  * http://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 /**
  28  * @ingroup Dump Maintenance
  29  */
  30 class DumpDBZip2Output extends DumpPipeOutput {
  31         function DumpDBZip2Output( $file ) {
  32                 parent::__construct( "dbzip2", $file );
  33         }
  34 }
  35
  36 /**
  37  * @ingroup Dump Maintenance
  38  */
  39 class BackupDumper {
  40         var $reportingInterval = 100;
  41         var $reporting = true;
  42         var $pageCount = 0;
  43         var $revCount  = 0;
  44         var $server    = null; // use default
  45         var $pages     = null; // all pages
  46         var $skipHeader = false; // don't output <mediawiki> and <siteinfo>
  47         var $skipFooter = false; // don't output </mediawiki>
  48         var $startId    = 0;
  49         var $endId      = 0;
  50         var $sink       = null; // Output filters
  51         var $stubText   = false; // include rev_text_id instead of text; for 2-pass dump
  52         var $dumpUploads = false;
  53         var $dumpUploadFileContents = false;
  54         var $lastTime = 0;
  55         var $pageCountLast = 0;
  56         var $revCountLast = 0;
  57         var $ID = 0;
  58
  59         function BackupDumper( $args ) {
  60                 $this->stderr = fopen( "php://stderr", "wt" );
  61
  62                 // Built-in output and filter plugins
  63                 $this->registerOutput( 'file', 'DumpFileOutput' );
  64                 $this->registerOutput( 'gzip', 'DumpGZipOutput' );
  65                 $this->registerOutput( 'bzip2', 'DumpBZip2Output' );
  66                 $this->registerOutput( 'dbzip2', 'DumpDBZip2Output' );
  67                 $this->registerOutput( '7zip', 'Dump7ZipOutput' );
  68
  69                 $this->registerFilter( 'latest', 'DumpLatestFilter' );
  70                 $this->registerFilter( 'notalk', 'DumpNotalkFilter' );
  71                 $this->registerFilter( 'namespace', 'DumpNamespaceFilter' );
  72
  73                 $this->sink = $this->processArgs( $args );
  74         }
  75
  76         /**
  77          * @param $name String
  78          * @param $class String: name of output filter plugin class
  79          */
  80         function registerOutput( $name, $class ) {
  81                 $this->outputTypes[$name] = $class;
  82         }
  83
  84         /**
  85          * @param $name String
  86          * @param $class String: name of filter plugin class
  87          */
  88         function registerFilter( $name, $class ) {
  89                 $this->filterTypes[$name] = $class;
  90         }
  91
  92         /**
  93          * Load a plugin and register it
  94          *
  95          * @param $class String: name of plugin class; must have a static 'register'
  96          *               method that takes a BackupDumper as a parameter.
  97          * @param $file String: full or relative path to the PHP file to load, or empty
  98          */
  99         function loadPlugin( $class, $file ) {
 100                 if ( $file != '' ) {
 101                         require_once( $file );
 102                 }
 103                 $register = array( $class, 'register' );
 104                 call_user_func_array( $register, array( &$this ) );
 105         }
 106
 107         /**
 108          * @param $args Array
 109          * @return Array
 110          */
 111         function processArgs( $args ) {
 112                 $sink = null;
 113                 $sinks = array();
 114                 foreach ( $args as $arg ) {
 115                         $matches = array();
 116                         if ( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
 117                                 @list( /* $full */ , $opt, $val, $param ) = $matches;
 118                                 switch( $opt ) {
 119                                 case "plugin":
 120                                         $this->loadPlugin( $val, $param );
 121                                         break;
 122                                 case "output":
 123                                         if ( !is_null( $sink ) ) {
 124                                                 $sinks[] = $sink;
 125                                         }
 126                                         if ( !isset( $this->outputTypes[$val] ) ) {
 127                                                 $this->fatalError( "Unrecognized output sink type '$val'" );
 128                                         }
 129                                         $type = $this->outputTypes[$val];
 130                                         $sink = new $type( $param );
 131                                         break;
 132                                 case "filter":
 133                                         if ( is_null( $sink ) ) {
 134                                                 $sink = new DumpOutput();
 135                                         }
 136                                         if ( !isset( $this->filterTypes[$val] ) ) {
 137                                                 $this->fatalError( "Unrecognized filter type '$val'" );
 138                                         }
 139                                         $type = $this->filterTypes[$val];
 140                                         $filter = new $type( $sink, $param );
 141
 142                                         // references are lame in php...
 143                                         unset( $sink );
 144                                         $sink = $filter;
 145
 146                                         break;
 147                                 case "report":
 148                                         $this->reportingInterval = intval( $val );
 149                                         break;
 150                                 case "server":
 151                                         $this->server = $val;
 152                                         break;
 153                                 case "force-normal":
 154                                         if ( !function_exists( 'utf8_normalize' ) ) {
 155                                                 wfDl( "php_utfnormal.so" );
 156                                                 if ( !function_exists( 'utf8_normalize' ) ) {
 157                                                         $this->fatalError( "Failed to load UTF-8 normalization extension. " .
 158                                                                 "Install or remove --force-normal parameter to use slower code." );
 159                                                 }
 160                                         }
 161                                         break;
 162                                 default:
 163                                         $this->processOption( $opt, $val, $param );
 164                                 }
 165                         }
 166                 }
 167
 168                 if ( is_null( $sink ) ) {
 169                         $sink = new DumpOutput();
 170                 }
 171                 $sinks[] = $sink;
 172
 173                 if ( count( $sinks ) > 1 ) {
 174                         return new DumpMultiWriter( $sinks );
 175                 } else {
 176                         return $sink;
 177                 }
 178         }
 179
 180         function processOption( $opt, $val, $param ) {
 181                 // extension point for subclasses to add options
 182         }
 183
 184         function dump( $history, $text = WikiExporter::TEXT ) {
 185                 # Notice messages will foul up your XML output even if they're
 186                 # relatively harmless.
 187                 if ( ini_get( 'display_errors' ) )
 188                         ini_set( 'display_errors', 'stderr' );
 189
 190                 $this->initProgress( $history );
 191
 192                 $db = $this->backupDb();
 193                 $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
 194                 $exporter->dumpUploads = $this->dumpUploads;
 195                 $exporter->dumpUploadFileContents = $this->dumpUploadFileContents;
 196
 197                 $wrapper = new ExportProgressFilter( $this->sink, $this );
 198                 $exporter->setOutputSink( $wrapper );
 199
 200                 if ( !$this->skipHeader )
 201                         $exporter->openStream();
 202                 # Log item dumps: all or by range
 203                 if ( $history & WikiExporter::LOGS ) {
 204                         if ( $this->startId || $this->endId ) {
 205                                 $exporter->logsByRange( $this->startId, $this->endId );
 206                         } else {
 207                                 $exporter->allLogs();
 208                         }
 209                 # Page dumps: all or by page ID range
 210                 } else if ( is_null( $this->pages ) ) {
 211                         if ( $this->startId || $this->endId ) {
 212                                 $exporter->pagesByRange( $this->startId, $this->endId );
 213                         } else {
 214                                 $exporter->allPages();
 215                         }
 216                 # Dump of specific pages
 217                 } else {
 218                         $exporter->pagesByName( $this->pages );
 219                 }
 220
 221                 if ( !$this->skipFooter )
 222                         $exporter->closeStream();
 223
 224                 $this->report( true );
 225         }
 226
 227         /**
 228          * Initialise starting time and maximum revision count.
 229          * We'll make ETA calculations based an progress, assuming relatively
 230          * constant per-revision rate.
 231          * @param $history Integer: WikiExporter::CURRENT or WikiExporter::FULL
 232          */
 233         function initProgress( $history = WikiExporter::FULL ) {
 234                 $table = ( $history == WikiExporter::CURRENT ) ? 'page' : 'revision';
 235                 $field = ( $history == WikiExporter::CURRENT ) ? 'page_id' : 'rev_id';
 236
 237                 $dbr = wfGetDB( DB_SLAVE );
 238                 $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', __METHOD__ );
 239                 $this->startTime = wfTime();
 240                 $this->lastTime = $this->startTime;
 241                 $this->ID = getmypid();
 242         }
 243
 244         /**
 245          * @todo Fixme: the --server parameter is currently not respected, as it
 246          * doesn't seem terribly easy to ask the load balancer for a particular
 247          * connection by name.
 248          */
 249         function backupDb() {
 250                 $this->lb = wfGetLBFactory()->newMainLB();
 251                 $db = $this->lb->getConnection( DB_SLAVE, 'backup' );
 252
 253                 // Discourage the server from disconnecting us if it takes a long time
 254                 // to read out the big ol' batch query.
 255                 $db->setTimeout( 3600 * 24 );
 256
 257                 return $db;
 258         }
 259
 260         function __destruct() {
 261                 if ( isset( $this->lb ) ) {
 262                         $this->lb->closeAll();
 263                 }
 264         }
 265
 266         function backupServer() {
 267                 global $wgDBserver;
 268                 return $this->server
 269                         ? $this->server
 270                         : $wgDBserver;
 271         }
 272
 273         function reportPage() {
 274                 $this->pageCount++;
 275         }
 276
 277         function revCount() {
 278                 $this->revCount++;
 279                 $this->report();
 280         }
 281
 282         function report( $final = false ) {
 283                 if ( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
 284                         $this->showReport();
 285                 }
 286         }
 287
 288         function showReport() {
 289                 if ( $this->reporting ) {
 290                         $now = wfTimestamp( TS_DB );
 291                         $nowts = wfTime();
 292                         $deltaAll = wfTime() - $this->startTime;
 293                         $deltaPart = wfTime() - $this->lastTime;
 294                         $this->pageCountPart = $this->pageCount - $this->pageCountLast;
 295                         $this->revCountPart = $this->revCount - $this->revCountLast;
 296
 297                         if ( $deltaAll ) {
 298                                 $portion = $this->revCount / $this->maxCount;
 299                                 $eta = $this->startTime + $deltaAll / $portion;
 300                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 301                                 $pageRate = $this->pageCount / $deltaAll;
 302                                 $revRate = $this->revCount / $deltaAll;
 303                         } else {
 304                                 $pageRate = '-';
 305                                 $revRate = '-';
 306                                 $etats = '-';
 307                         }
 308                         if ( $deltaPart ) {
 309                                 $pageRatePart = $this->pageCountPart / $deltaPart;
 310                                 $revRatePart = $this->revCountPart / $deltaPart;
 311                         } else {
 312                                 $pageRatePart = '-';
 313                                 $revRatePart = '-';
 314                         }
 315                         $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]",
 316                                         $now, wfWikiID(), $this->ID, $this->pageCount, $pageRate, $pageRatePart, $this->revCount, $revRate, $revRatePart, $etats, $this->maxCount ) );
 317                         $this->lastTime = $nowts;
 318                         $this->revCountLast = $this->revCount;
 319                 }
 320         }
 321
 322         function progress( $string ) {
 323                 fwrite( $this->stderr, $string . "\n" );
 324         }
 325
 326         function fatalError( $msg ) {
 327                 $this->progress( "$msg\n" );
 328                 die(1);
 329         }
 330 }
 331
 332 class ExportProgressFilter extends DumpFilter {
 333         function ExportProgressFilter( &$sink, &$progress ) {
 334                 parent::__construct( $sink );
 335                 $this->progress = $progress;
 336         }
 337
 338         function writeClosePage( $string ) {
 339                 parent::writeClosePage( $string );
 340                 $this->progress->reportPage();
 341         }
 342
 343         function writeRevision( $rev, $string ) {
 344                 parent::writeRevision( $rev, $string );
 345                 $this->progress->revCount();
 346         }
 347 }