maintenance/orphans.php

   1 <?php
   2 # Copyright (C) 2005 Brion Vibber <brion@pobox.com>
   3 # http://www.mediawiki.org/
   4 #
   5 # This program is free software; you can redistribute it and/or modify
   6 # it under the terms of the GNU General Public License as published by
   7 # the Free Software Foundation; either version 2 of the License, or
   8 # (at your option) any later version.
   9 #
  10 # This program is distributed in the hope that it will be useful,
  11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13 # GNU General Public License for more details.
  14 #
  15 # You should have received a copy of the GNU General Public License along
  16 # with this program; if not, write to the Free Software Foundation, Inc.,
  17 # 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18 # http://www.gnu.org/copyleft/gpl.html
  19
  20 /**
  21  * Look for 'orphan' revisions hooked to pages which don't exist
  22  * And 'childless' pages with no revisions.
  23  * Then, kill the poor widows and orphans.
  24  * Man this is depressing.
  25  *
  26  * @file
  27  * @author <brion@pobox.com>
  28  * @ingroup Maintenance
  29  */
  30
  31 $options = array( 'fix' );
  32
  33 /** */
  34 require_once( 'commandLine.inc' );
  35 $wgTitle = Title::newFromText( 'Orphan revision cleanup script' );
  36
  37 checkOrphans( isset( $options['fix'] ) );
  38 checkSeparation( isset( $options['fix'] ) );
  39 #checkWidows( isset( $options['fix'] ) );
  40
  41 # ------
  42
  43 function checkOrphans( $fix ) {
  44         $dbw = wfGetDB( DB_MASTER );
  45         $page = $dbw->tableName( 'page' );
  46         $revision = $dbw->tableName( 'revision' );
  47
  48         if( $fix ) {
  49                 $dbw->query( "LOCK TABLES $page WRITE, $revision WRITE" );
  50         }
  51
  52         echo "Checking for orphan revision table entries... (this may take a while on a large wiki)\n";
  53         $result = $dbw->query( "
  54                 SELECT *
  55                 FROM $revision LEFT OUTER JOIN $page ON rev_page=page_id
  56                 WHERE page_id IS NULL
  57         ");
  58         $orphans = $dbw->numRows( $result );
  59         if( $orphans > 0 ) {
  60                 global $wgContLang;
  61                 echo "$orphans orphan revisions...\n";
  62                 printf( "%10s %10s %14s %20s %s\n", 'rev_id', 'rev_page', 'rev_timestamp', 'rev_user_text', 'rev_comment' );
  63                 while( $row = $dbw->fetchObject( $result ) ) {
  64                         $comment = ( $row->rev_comment == '' )
  65                                 ? ''
  66                                 : '(' . $wgContLang->truncate( $row->rev_comment, 40, '...' ) . ')';
  67                         printf( "%10d %10d %14s %20s %s\n",
  68                                 $row->rev_id,
  69                                 $row->rev_page,
  70                                 $row->rev_timestamp,
  71                                 $wgContLang->truncate( $row->rev_user_text, 17, '...' ),
  72                                 $comment );
  73                         if( $fix ) {
  74                                 $dbw->delete( 'revision', array( 'rev_id' => $row->rev_id ) );
  75                         }
  76                 }
  77                 if( !$fix ) {
  78                         echo "Run again with --fix to remove these entries automatically.\n";
  79                 }
  80         } else {
  81                 echo "No orphans! Yay!\n";
  82         }
  83
  84         if( $fix ) {
  85                 $dbw->query( "UNLOCK TABLES" );
  86         }
  87 }
  88
  89 /**
  90  * @todo DON'T USE THIS YET! It will remove entries which have children,
  91  *       but which aren't properly attached (eg if page_latest is bogus
  92  *       but valid revisions do exist)
  93  */
  94 function checkWidows( $fix ) {
  95         $dbw = wfGetDB( DB_MASTER );
  96         $page = $dbw->tableName( 'page' );
  97         $revision = $dbw->tableName( 'revision' );
  98
  99         if( $fix ) {
 100                 $dbw->query( "LOCK TABLES $page WRITE, $revision WRITE" );
 101         }
 102
 103         echo "\nChecking for childless page table entries... (this may take a while on a large wiki)\n";
 104         $result = $dbw->query( "
 105                 SELECT *
 106                 FROM $page LEFT OUTER JOIN $revision ON page_latest=rev_id
 107                 WHERE rev_id IS NULL
 108         ");
 109         $widows = $dbw->numRows( $result );
 110         if( $widows > 0 ) {
 111                 global $wgContLang;
 112                 echo "$widows childless pages...\n";
 113                 printf( "%10s %11s %2s %s\n", 'page_id', 'page_latest', 'ns', 'page_title' );
 114                 while( $row = $dbw->fetchObject( $result ) ) {
 115                         printf( "%10d %11d %2d %s\n",
 116                                 $row->page_id,
 117                                 $row->page_latest,
 118                                 $row->page_namespace,
 119                                 $row->page_title );
 120                         if( $fix ) {
 121                                 $dbw->delete( 'page', array( 'page_id' => $row->page_id ) );
 122                         }
 123                 }
 124                 if( !$fix ) {
 125                         echo "Run again with --fix to remove these entries automatically.\n";
 126                 }
 127         } else {
 128                 echo "No childless pages! Yay!\n";
 129         }
 130
 131         if( $fix ) {
 132                 $dbw->query( "UNLOCK TABLES" );
 133         }
 134 }
 135
 136
 137 function checkSeparation( $fix ) {
 138         $dbw = wfGetDB( DB_MASTER );
 139         $page     = $dbw->tableName( 'page' );
 140         $revision = $dbw->tableName( 'revision' );
 141         $text     = $dbw->tableName( 'text' );
 142
 143         if( $fix ) {
 144                 $dbw->query( "LOCK TABLES $page WRITE, $revision WRITE, $text WRITE" );
 145         }
 146
 147         echo "\nChecking for pages whose page_latest links are incorrect... (this may take a while on a large wiki)\n";
 148         $result = $dbw->query( "
 149                 SELECT *
 150                 FROM $page LEFT OUTER JOIN $revision ON page_latest=rev_id
 151         ");
 152         $found = 0;
 153         while( $row = $dbw->fetchObject( $result ) ) {
 154                 $result2 = $dbw->query( "
 155                         SELECT MAX(rev_timestamp) as max_timestamp
 156                         FROM $revision
 157                         WHERE rev_page=$row->page_id
 158                 " );
 159                 $row2 = $dbw->fetchObject( $result2 );
 160                 $dbw->freeResult( $result2 );
 161                 if( $row2 ) {
 162                         if( $row->rev_timestamp != $row2->max_timestamp ) {
 163                                 if( $found == 0 ) {
 164                                         printf( "%10s %10s %14s %14s\n",
 165                                                 'page_id', 'rev_id', 'timestamp', 'max timestamp' );
 166                                 }
 167                                 ++$found;
 168                                 printf( "%10d %10d %14s %14s\n",
 169                                         $row->page_id,
 170                                         $row->page_latest,
 171                                         $row->rev_timestamp,
 172                                         $row2->max_timestamp );
 173                                 if( $fix ) {
 174                                         # ...
 175                                         $maxId = $dbw->selectField(
 176                                                 'revision',
 177                                                 'rev_id',
 178                                                 array(
 179                                                         'rev_page'      => $row->page_id,
 180                                                         'rev_timestamp' => $row2->max_timestamp ) );
 181                                         echo "... updating to revision $maxId\n";
 182                                         $maxRev = Revision::newFromId( $maxId );
 183                                         $title = Title::makeTitle( $row->page_namespace, $row->page_title );
 184                                         $article = new Article( $title );
 185                                         $article->updateRevisionOn( $dbw, $maxRev );
 186                                 }
 187                         }
 188                 } else {
 189                         echo "wtf\n";
 190                 }
 191         }
 192
 193         if( $found ) {
 194                 echo "Found $found pages with incorrect latest revision.\n";
 195         } else {
 196                 echo "No pages with incorrect latest revision. Yay!\n";
 197         }
 198         if( !$fix && $found > 0 ) {
 199                 echo "Run again with --fix to remove these entries automatically.\n";
 200         }
 201
 202         if( $fix ) {
 203                 $dbw->query( "UNLOCK TABLES" );
 204         }
 205 }
 206