maintenance/backup.inc

   1 <?php
   2 /**
   3  * Copyright (C) 2005 Brion Vibber <brion@pobox.com>
   4  * http://www.mediawiki.org/
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License along
  17  * with this program; if not, write to the Free Software Foundation, Inc.,
  18  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  19  * http://www.gnu.org/copyleft/gpl.html
  20  *
  21  * @addtogroup SpecialPage
  22  */
  23
  24 class DumpDBZip2Output extends DumpPipeOutput {
  25         function DumpDBZip2Output( $file ) {
  26                 parent::DumpPipeOutput( "dbzip2", $file );
  27         }
  28 }
  29
  30 class BackupDumper {
  31         var $reportingInterval = 100;
  32         var $reporting = true;
  33         var $pageCount = 0;
  34         var $revCount  = 0;
  35         var $server    = null; // use default
  36         var $pages     = null; // all pages
  37         var $skipHeader = false; // don't output <mediawiki> and <siteinfo>
  38         var $skipFooter = false; // don't output </mediawiki>
  39         var $startId    = 0;
  40         var $endId      = 0;
  41         var $sink       = null; // Output filters
  42         var $stubText   = false; // include rev_text_id instead of text; for 2-pass dump
  43         var $dumpUploads = false;
  44
  45         function BackupDumper( $args ) {
  46                 $this->stderr = fopen( "php://stderr", "wt" );
  47
  48                 // Built-in output and filter plugins
  49                 $this->registerOutput( 'file', 'DumpFileOutput' );
  50                 $this->registerOutput( 'gzip', 'DumpGZipOutput' );
  51                 $this->registerOutput( 'bzip2', 'DumpBZip2Output' );
  52                 $this->registerOutput( 'dbzip2', 'DumpDBZip2Output' );
  53                 $this->registerOutput( '7zip', 'Dump7ZipOutput' );
  54
  55                 $this->registerFilter( 'latest', 'DumpLatestFilter' );
  56                 $this->registerFilter( 'notalk', 'DumpNotalkFilter' );
  57                 $this->registerFilter( 'namespace', 'DumpNamespaceFilter' );
  58
  59                 $this->sink = $this->processArgs( $args );
  60         }
  61
  62         /**
  63          * @param string $name
  64          * @param string $class name of output filter plugin class
  65          */
  66         function registerOutput( $name, $class ) {
  67                 $this->outputTypes[$name] = $class;
  68         }
  69
  70         /**
  71          * @param string $name
  72          * @param string $class name of filter plugin class
  73          */
  74         function registerFilter( $name, $class ) {
  75                 $this->filterTypes[$name] = $class;
  76         }
  77
  78         /**
  79          * Load a plugin and register it
  80          * @param string $class Name of plugin class; must have a static 'register'
  81          *                      method that takes a BackupDumper as a parameter.
  82          * @param string $file Full or relative path to the PHP file to load, or empty
  83          */
  84         function loadPlugin( $class, $file ) {
  85                 if( $file != '' ) {
  86                         require_once( $file );
  87                 }
  88                 $register = array( $class, 'register' );
  89                 call_user_func_array( $register, array( &$this ) );
  90         }
  91
  92         /**
  93          * @param array $args
  94          * @return array
  95          * @static
  96          */
  97         function processArgs( $args ) {
  98                 $sink = null;
  99                 $sinks = array();
 100                 foreach( $args as $arg ) {
 101                         $matches = array();
 102                         if( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
 103                                 @list( /* $full */ , $opt, $val, $param ) = $matches;
 104                                 switch( $opt ) {
 105                                 case "plugin":
 106                                         $this->loadPlugin( $val, $param );
 107                                         break;
 108                                 case "output":
 109                                         if( !is_null( $sink ) ) {
 110                                                 $sinks[] = $sink;
 111                                         }
 112                                         if( !isset( $this->outputTypes[$val] ) ) {
 113                                                 wfDie( "Unrecognized output sink type '$val'\n" );
 114                                         }
 115                                         $type = $this->outputTypes[$val];
 116                                         $sink = new $type( $param );
 117                                         break;
 118                                 case "filter":
 119                                         if( is_null( $sink ) ) {
 120                                                 $this->progress( "Warning: assuming stdout for filter output\n" );
 121                                                 $sink = new DumpOutput();
 122                                         }
 123                                         if( !isset( $this->filterTypes[$val] ) ) {
 124                                                 wfDie( "Unrecognized filter type '$val'\n" );
 125                                         }
 126                                         $type = $this->filterTypes[$val];
 127                                         $filter = new $type( $sink, $param );
 128
 129                                         // references are lame in php...
 130                                         unset( $sink );
 131                                         $sink = $filter;
 132
 133                                         break;
 134                                 case "report":
 135                                         $this->reportingInterval = intval( $val );
 136                                         break;
 137                                 case "server":
 138                                         $this->server = $val;
 139                                         break;
 140                                 case "force-normal":
 141                                         if( !function_exists( 'utf8_normalize' ) ) {
 142                                                 dl( "php_utfnormal.so" );
 143                                                 if( !function_exists( 'utf8_normalize' ) ) {
 144                                                         wfDie( "Failed to load UTF-8 normalization extension. " .
 145                                                                 "Install or remove --force-normal parameter to use slower code.\n" );
 146                                                 }
 147                                         }
 148                                         break;
 149                                 default:
 150                                         $this->processOption( $opt, $val, $param );
 151                                 }
 152                         }
 153                 }
 154
 155                 if( is_null( $sink ) ) {
 156                         $sink = new DumpOutput();
 157                 }
 158                 $sinks[] = $sink;
 159
 160                 if( count( $sinks ) > 1 ) {
 161                         return new DumpMultiWriter( $sinks );
 162                 } else {
 163                         return $sink;
 164                 }
 165         }
 166
 167         function processOption( $opt, $val, $param ) {
 168                 // extension point for subclasses to add options
 169         }
 170
 171         function dump( $history, $text = MW_EXPORT_TEXT ) {
 172                 # Notice messages will foul up your XML output even if they're
 173                 # relatively harmless.
 174                 if( ini_get( 'display_errors' ) )
 175                         ini_set( 'display_errors', 'stderr' );
 176
 177                 $this->initProgress( $history );
 178
 179                 $db = $this->backupDb();
 180                 $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
 181                 $exporter->dumpUploads = $this->dumpUploads;
 182
 183                 $wrapper = new ExportProgressFilter( $this->sink, $this );
 184                 $exporter->setOutputSink( $wrapper );
 185
 186                 if( !$this->skipHeader )
 187                         $exporter->openStream();
 188
 189                 if( is_null( $this->pages ) ) {
 190                         if( $this->startId || $this->endId ) {
 191                                 $exporter->pagesByRange( $this->startId, $this->endId );
 192                         } else {
 193                                 $exporter->allPages();
 194                         }
 195                 } else {
 196                         $exporter->pagesByName( $this->pages );
 197                 }
 198
 199                 if( !$this->skipFooter )
 200                         $exporter->closeStream();
 201
 202                 $this->report( true );
 203         }
 204
 205         /**
 206          * Initialise starting time and maximum revision count.
 207          * We'll make ETA calculations based an progress, assuming relatively
 208          * constant per-revision rate.
 209          * @param int $history WikiExporter::CURRENT or WikiExporter::FULL
 210          */
 211         function initProgress( $history = WikiExporter::FULL ) {
 212                 $table = ($history == WikiExporter::CURRENT) ? 'page' : 'revision';
 213                 $field = ($history == WikiExporter::CURRENT) ? 'page_id' : 'rev_id';
 214
 215                 $dbr = wfGetDB( DB_SLAVE );
 216                 $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', 'BackupDumper::dump' );
 217                 $this->startTime = wfTime();
 218         }
 219
 220         function backupDb() {
 221                 global $wgDBadminuser, $wgDBadminpassword;
 222                 global $wgDBname, $wgDebugDumpSql, $wgDBtype;
 223                 $flags = ($wgDebugDumpSql ? DBO_DEBUG : 0) | DBO_DEFAULT; // god-damn hack
 224
 225                 $class = 'Database' . ucfirst($wgDBtype);
 226                 $db = new $class( $this->backupServer(), $wgDBadminuser, $wgDBadminpassword, $wgDBname, false, $flags );
 227
 228                 // Discourage the server from disconnecting us if it takes a long time
 229                 // to read out the big ol' batch query.
 230                 $db->setTimeout( 3600 * 24 );
 231
 232                 return $db;
 233         }
 234
 235         function backupServer() {
 236                 global $wgDBserver;
 237                 return $this->server
 238                         ? $this->server
 239                         : $wgDBserver;
 240         }
 241
 242         function reportPage() {
 243                 $this->pageCount++;
 244         }
 245
 246         function revCount() {
 247                 $this->revCount++;
 248                 $this->report();
 249         }
 250
 251         function report( $final = false ) {
 252                 if( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
 253                         $this->showReport();
 254                 }
 255         }
 256
 257         function showReport() {
 258                 if( $this->reporting ) {
 259                         $delta = wfTime() - $this->startTime;
 260                         $now = wfTimestamp( TS_DB );
 261                         if( $delta ) {
 262                                 $rate = $this->pageCount / $delta;
 263                                 $revrate = $this->revCount / $delta;
 264                                 $portion = $this->revCount / $this->maxCount;
 265                                 $eta = $this->startTime + $delta / $portion;
 266                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 267                         } else {
 268                                 $rate = '-';
 269                                 $revrate = '-';
 270                                 $etats = '-';
 271                         }
 272                         $this->progress( sprintf( "%s: %s %d pages (%0.3f/sec), %d revs (%0.3f/sec), ETA %s [max %d]",
 273                                 $now, wfWikiID(), $this->pageCount, $rate, $this->revCount, $revrate, $etats, $this->maxCount ) );
 274                 }
 275         }
 276
 277         function progress( $string ) {
 278                 fwrite( $this->stderr, $string . "\n" );
 279         }
 280 }
 281
 282 class ExportProgressFilter extends DumpFilter {
 283         function ExportProgressFilter( &$sink, &$progress ) {
 284                 parent::DumpFilter( $sink );
 285                 $this->progress = $progress;
 286         }
 287
 288         function writeClosePage( $string ) {
 289                 parent::writeClosePage( $string );
 290                 $this->progress->reportPage();
 291         }
 292
 293         function writeRevision( $rev, $string ) {
 294                 parent::writeRevision( $rev, $string );
 295                 $this->progress->revCount();
 296         }
 297 }
 298
 299 ?>