maintenance/importDump.php

   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 Maintenance
  23  */
  24
  25 $optionsWithArgs = array( 'report' );
  26
  27 require_once( dirname( __FILE__ ) . '/commandLine.inc' );
  28
  29 /**
  30  * @ingroup Maintenance
  31  */
  32 class BackupReader {
  33         var $reportingInterval = 100;
  34         var $reporting = true;
  35         var $pageCount = 0;
  36         var $revCount  = 0;
  37         var $dryRun    = false;
  38         var $debug     = false;
  39         var $uploads   = false;
  40
  41         function __construct() {
  42                 $this->stderr = fopen( "php://stderr", "wt" );
  43         }
  44
  45         function reportPage( $page ) {
  46                 $this->pageCount++;
  47         }
  48
  49         function handleRevision( $rev ) {
  50                 $title = $rev->getTitle();
  51                 if ( !$title ) {
  52                         $this->progress( "Got bogus revision with null title!" );
  53                         return;
  54                 }
  55
  56                 $this->revCount++;
  57                 $this->report();
  58
  59                 if ( !$this->dryRun ) {
  60                         call_user_func( $this->importCallback, $rev );
  61                 }
  62         }
  63
  64         function handleUpload( $revision ) {
  65                 if ( $this->uploads ) {
  66                         $this->uploadCount++;
  67                         // $this->report();
  68                         $this->progress( "upload: " . $revision->getFilename() );
  69
  70                         if ( !$this->dryRun ) {
  71                                 // bluuuh hack
  72                                 // call_user_func( $this->uploadCallback, $revision );
  73                                 $dbw = wfGetDB( DB_MASTER );
  74                                 return $dbw->deadlockLoop( array( $revision, 'importUpload' ) );
  75                         }
  76                 }
  77         }
  78
  79         function handleLogItem( $rev ) {
  80                 $this->revCount++;
  81                 $this->report();
  82
  83                 if ( !$this->dryRun ) {
  84                         call_user_func( $this->logItemCallback, $rev );
  85                 }
  86         }
  87
  88         function report( $final = false ) {
  89                 if ( $final xor ( $this->pageCount % $this->reportingInterval == 0 ) ) {
  90                         $this->showReport();
  91                 }
  92         }
  93
  94         function showReport() {
  95                 if ( $this->reporting ) {
  96                         $delta = wfTime() - $this->startTime;
  97                         if ( $delta ) {
  98                                 $rate = sprintf( "%.2f", $this->pageCount / $delta );
  99                                 $revrate = sprintf( "%.2f", $this->revCount / $delta );
 100                         } else {
 101                                 $rate = '-';
 102                                 $revrate = '-';
 103                         }
 104                         # Logs dumps don't have page tallies
 105                         if ( $this->pageCount )
 106                                 $this->progress( "$this->pageCount ($rate pages/sec $revrate revs/sec)" );
 107                         else
 108                                 $this->progress( "$this->revCount ($revrate revs/sec)" );
 109                 }
 110                 wfWaitForSlaves( 5 );
 111         }
 112
 113         function progress( $string ) {
 114                 fwrite( $this->stderr, $string . "\n" );
 115         }
 116
 117         function importFromFile( $filename ) {
 118                 if ( preg_match( '/\.gz$/', $filename ) ) {
 119                         $filename = 'compress.zlib://' . $filename;
 120                 }
 121                 elseif ( preg_match( '/\.bz2$/', $filename ) ) {
 122                         $filename = 'compress.bzip2://' . $filename;
 123                 }
 124                 elseif ( preg_match( '/\.7z$/', $filename ) ) {
 125                         $filename = 'mediawiki.compress.7z://' . $filename;
 126                 }
 127
 128                 $file = fopen( $filename, 'rt' );
 129                 return $this->importFromHandle( $file );
 130         }
 131
 132         function importFromStdin() {
 133                 $file = fopen( 'php://stdin', 'rt' );
 134                 return $this->importFromHandle( $file );
 135         }
 136
 137         function importFromHandle( $handle ) {
 138                 $this->startTime = wfTime();
 139
 140                 $source = new ImportStreamSource( $handle );
 141                 $importer = new WikiImporter( $source );
 142
 143                 $importer->setDebug( $this->debug );
 144                 $importer->setPageCallback( array( &$this, 'reportPage' ) );
 145                 $this->importCallback =  $importer->setRevisionCallback(
 146                         array( &$this, 'handleRevision' ) );
 147                 $this->uploadCallback = $importer->setUploadCallback(
 148                         array( &$this, 'handleUpload' ) );
 149                 $this->logItemCallback = $importer->setLogItemCallback(
 150                         array( &$this, 'handleLogItem' ) );
 151
 152                 if ( $this->dryRun ) {
 153                         $importer->setPageOutCallback( null );
 154                 }
 155
 156                 return $importer->doImport();
 157         }
 158 }
 159
 160 if ( wfReadOnly() ) {
 161         wfDie( "Wiki is in read-only mode; you'll need to disable it for import to work.\n" );
 162 }
 163
 164 $reader = new BackupReader();
 165 if ( isset( $options['quiet'] ) ) {
 166         $reader->reporting = false;
 167 }
 168 if ( isset( $options['report'] ) ) {
 169         $reader->reportingInterval = intval( $options['report'] );
 170 }
 171 if ( isset( $options['dry-run'] ) ) {
 172         $reader->dryRun = true;
 173 }
 174 if ( isset( $options['debug'] ) ) {
 175         $reader->debug = true;
 176 }
 177 if ( isset( $options['uploads'] ) ) {
 178         $reader->uploads = true; // experimental!
 179 }
 180
 181 if ( isset( $args[0] ) ) {
 182         $result = $reader->importFromFile( $args[0] );
 183 } else {
 184         $result = $reader->importFromStdin();
 185 }
 186
 187 echo "Done!\n";
 188 echo "You might want to run rebuildrecentchanges.php to regenerate\n";
 189 echo "the recentchanges page.\n";