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  * @file
  22  * @ingroup Dump Maintenance
  23  */
  24
  25 /**
  26  * @ingroup Dump Maintenance
  27  */
  28 class DumpDBZip2Output extends DumpPipeOutput {
  29         function DumpDBZip2Output( $file ) {
  30                 parent::DumpPipeOutput( "dbzip2", $file );
  31         }
  32 }
  33
  34 /**
  35  * @ingroup Dump Maintenance
  36  */
  37 class BackupDumper {
  38         var $reportingInterval = 100;
  39         var $reporting = true;
  40         var $pageCount = 0;
  41         var $revCount  = 0;
  42         var $server    = null; // use default
  43         var $pages     = null; // all pages
  44         var $skipHeader = false; // don't output <mediawiki> and <siteinfo>
  45         var $skipFooter = false; // don't output </mediawiki>
  46         var $startId    = 0;
  47         var $endId      = 0;
  48         var $sink       = null; // Output filters
  49         var $stubText   = false; // include rev_text_id instead of text; for 2-pass dump
  50         var $dumpUploads = false;
  51
  52         function BackupDumper( $args ) {
  53                 $this->stderr = fopen( "php://stderr", "wt" );
  54
  55                 // Built-in output and filter plugins
  56                 $this->registerOutput( 'file', 'DumpFileOutput' );
  57                 $this->registerOutput( 'gzip', 'DumpGZipOutput' );
  58                 $this->registerOutput( 'bzip2', 'DumpBZip2Output' );
  59                 $this->registerOutput( 'dbzip2', 'DumpDBZip2Output' );
  60                 $this->registerOutput( '7zip', 'Dump7ZipOutput' );
  61
  62                 $this->registerFilter( 'latest', 'DumpLatestFilter' );
  63                 $this->registerFilter( 'notalk', 'DumpNotalkFilter' );
  64                 $this->registerFilter( 'namespace', 'DumpNamespaceFilter' );
  65
  66                 $this->sink = $this->processArgs( $args );
  67         }
  68
  69         /**
  70          * @param string $name
  71          * @param string $class name of output filter plugin class
  72          */
  73         function registerOutput( $name, $class ) {
  74                 $this->outputTypes[$name] = $class;
  75         }
  76
  77         /**
  78          * @param string $name
  79          * @param string $class name of filter plugin class
  80          */
  81         function registerFilter( $name, $class ) {
  82                 $this->filterTypes[$name] = $class;
  83         }
  84
  85         /**
  86          * Load a plugin and register it
  87          * @param string $class Name of plugin class; must have a static 'register'
  88          *                      method that takes a BackupDumper as a parameter.
  89          * @param string $file Full or relative path to the PHP file to load, or empty
  90          */
  91         function loadPlugin( $class, $file ) {
  92                 if( $file != '' ) {
  93                         require_once( $file );
  94                 }
  95                 $register = array( $class, 'register' );
  96                 call_user_func_array( $register, array( &$this ) );
  97         }
  98
  99         /**
 100          * @param array $args
 101          * @return array
 102          * @static
 103          */
 104         function processArgs( $args ) {
 105                 $sink = null;
 106                 $sinks = array();
 107                 foreach( $args as $arg ) {
 108                         $matches = array();
 109                         if( preg_match( '/^--(.+?)(?:=(.+?)(?::(.+?))?)?$/', $arg, $matches ) ) {
 110                                 @list( /* $full */ , $opt, $val, $param ) = $matches;
 111                                 switch( $opt ) {
 112                                 case "plugin":
 113                                         $this->loadPlugin( $val, $param );
 114                                         break;
 115                                 case "output":
 116                                         if( !is_null( $sink ) ) {
 117                                                 $sinks[] = $sink;
 118                                         }
 119                                         if( !isset( $this->outputTypes[$val] ) ) {
 120                                                 wfDie( "Unrecognized output sink type '$val'\n" );
 121                                         }
 122                                         $type = $this->outputTypes[$val];
 123                                         $sink = new $type( $param );
 124                                         break;
 125                                 case "filter":
 126                                         if( is_null( $sink ) ) {
 127                                                 $this->progress( "Warning: assuming stdout for filter output\n" );
 128                                                 $sink = new DumpOutput();
 129                                         }
 130                                         if( !isset( $this->filterTypes[$val] ) ) {
 131                                                 wfDie( "Unrecognized filter type '$val'\n" );
 132                                         }
 133                                         $type = $this->filterTypes[$val];
 134                                         $filter = new $type( $sink, $param );
 135
 136                                         // references are lame in php...
 137                                         unset( $sink );
 138                                         $sink = $filter;
 139
 140                                         break;
 141                                 case "report":
 142                                         $this->reportingInterval = intval( $val );
 143                                         break;
 144                                 case "server":
 145                                         $this->server = $val;
 146                                         break;
 147                                 case "force-normal":
 148                                         if( !function_exists( 'utf8_normalize' ) ) {
 149                                                 dl( "php_utfnormal.so" );
 150                                                 if( !function_exists( 'utf8_normalize' ) ) {
 151                                                         wfDie( "Failed to load UTF-8 normalization extension. " .
 152                                                                 "Install or remove --force-normal parameter to use slower code.\n" );
 153                                                 }
 154                                         }
 155                                         break;
 156                                 default:
 157                                         $this->processOption( $opt, $val, $param );
 158                                 }
 159                         }
 160                 }
 161
 162                 if( is_null( $sink ) ) {
 163                         $sink = new DumpOutput();
 164                 }
 165                 $sinks[] = $sink;
 166
 167                 if( count( $sinks ) > 1 ) {
 168                         return new DumpMultiWriter( $sinks );
 169                 } else {
 170                         return $sink;
 171                 }
 172         }
 173
 174         function processOption( $opt, $val, $param ) {
 175                 // extension point for subclasses to add options
 176         }
 177
 178         function dump( $history, $text = WikiExporter::TEXT ) {
 179                 # Notice messages will foul up your XML output even if they're
 180                 # relatively harmless.
 181                 if( ini_get( 'display_errors' ) )
 182                         ini_set( 'display_errors', 'stderr' );
 183
 184                 $this->initProgress( $history );
 185
 186                 $db = $this->backupDb();
 187                 $exporter = new WikiExporter( $db, $history, WikiExporter::STREAM, $text );
 188                 $exporter->dumpUploads = $this->dumpUploads;
 189
 190                 $wrapper = new ExportProgressFilter( $this->sink, $this );
 191                 $exporter->setOutputSink( $wrapper );
 192
 193                 if( !$this->skipHeader )
 194                         $exporter->openStream();
 195                 # Log item dumps: all or by range
 196                 if( $history & WikiExporter::LOGS ) {
 197                         if( $this->startId || $this->endId ) {
 198                                 $exporter->logsByRange( $this->startId, $this->endId );
 199                         } else {
 200                                 $exporter->allLogs();
 201                         }
 202                 # Page dumps: all or by page ID range
 203                 } else if( is_null( $this->pages ) ) {
 204                         if( $this->startId || $this->endId ) {
 205                                 $exporter->pagesByRange( $this->startId, $this->endId );
 206                         } else {
 207                                 $exporter->allPages();
 208                         }
 209                 # Dump of specific pages
 210                 } else {
 211                         $exporter->pagesByName( $this->pages );
 212                 }
 213
 214                 if( !$this->skipFooter )
 215                         $exporter->closeStream();
 216
 217                 $this->report( true );
 218         }
 219
 220         /**
 221          * Initialise starting time and maximum revision count.
 222          * We'll make ETA calculations based an progress, assuming relatively
 223          * constant per-revision rate.
 224          * @param int $history WikiExporter::CURRENT or WikiExporter::FULL
 225          */
 226         function initProgress( $history = WikiExporter::FULL ) {
 227                 $table = ($history == WikiExporter::CURRENT) ? 'page' : 'revision';
 228                 $field = ($history == WikiExporter::CURRENT) ? 'page_id' : 'rev_id';
 229
 230                 $dbr = wfGetDB( DB_SLAVE );
 231                 $this->maxCount = $dbr->selectField( $table, "MAX($field)", '', 'BackupDumper::dump' );
 232                 $this->startTime = wfTime();
 233         }
 234
 235         /**
 236          * @fixme the --server parameter is currently not respected, as it doesn't seem
 237          * terribly easy to ask the load balancer for a particular connection by name.
 238          */
 239         function backupDb() {
 240                 $this->lb = wfGetLBFactory()->newMainLB();
 241                 $db = $this->lb->getConnection( DB_SLAVE, 'backup' );
 242
 243                 // Discourage the server from disconnecting us if it takes a long time
 244                 // to read out the big ol' batch query.
 245                 $db->setTimeout( 3600 * 24 );
 246
 247                 return $db;
 248         }
 249
 250         function __destruct() {
 251                 if( isset( $this->lb ) ) {
 252                         $this->lb->closeAll();
 253                 }
 254         }
 255
 256         function backupServer() {
 257                 global $wgDBserver;
 258                 return $this->server
 259                         ? $this->server
 260                         : $wgDBserver;
 261         }
 262
 263         function reportPage() {
 264                 $this->pageCount++;
 265         }
 266
 267         function revCount() {
 268                 $this->revCount++;
 269                 $this->report();
 270         }
 271
 272         function report( $final = false ) {
 273                 if( $final xor ( $this->revCount % $this->reportingInterval == 0 ) ) {
 274                         $this->showReport();
 275                 }
 276         }
 277
 278         function showReport() {
 279                 if( $this->reporting ) {
 280                         $delta = wfTime() - $this->startTime;
 281                         $now = wfTimestamp( TS_DB );
 282                         if( $delta ) {
 283                                 $rate = $this->pageCount / $delta;
 284                                 $revrate = $this->revCount / $delta;
 285                                 $portion = $this->revCount / $this->maxCount;
 286                                 $eta = $this->startTime + $delta / $portion;
 287                                 $etats = wfTimestamp( TS_DB, intval( $eta ) );
 288                         } else {
 289                                 $rate = '-';
 290                                 $revrate = '-';
 291                                 $etats = '-';
 292                         }
 293                         $this->progress( sprintf( "%s: %s %d pages (%0.3f/sec), %d revs (%0.3f/sec), ETA %s [max %d]",
 294                                 $now, wfWikiID(), $this->pageCount, $rate, $this->revCount, $revrate, $etats, $this->maxCount ) );
 295                 }
 296         }
 297
 298         function progress( $string ) {
 299                 fwrite( $this->stderr, $string . "\n" );
 300         }
 301 }
 302
 303 class ExportProgressFilter extends DumpFilter {
 304         function ExportProgressFilter( &$sink, &$progress ) {
 305                 parent::DumpFilter( $sink );
 306                 $this->progress = $progress;
 307         }
 308
 309         function writeClosePage( $string ) {
 310                 parent::writeClosePage( $string );
 311                 $this->progress->reportPage();
 312         }
 313
 314         function writeRevision( $rev, $string ) {
 315                 parent::writeRevision( $rev, $string );
 316                 $this->progress->revCount();
 317         }
 318 }