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