maintenance/compressOld.inc

   1 <?php
   2 /**
   3  * @package MediaWiki
   4  * @subpackage Maintenance
   5  */
   6
   7 require_once( '../includes/Revision.php' );
   8
   9 /** */
  10 function compressOldPages( $start = 0 ) {
  11         $fname = 'compressOldPages';
  12
  13         $chunksize = 50;
  14         print "Starting from old_id $start...\n";
  15         $dbw =& wfGetDB( DB_MASTER );
  16         $old = $dbw->tableName( 'old' );
  17         do {
  18                 $end = $start + $chunksize;
  19                 $res = $dbw->select( 'old', array( 'old_id','old_flags','old_namespace','old_title','old_text' ),
  20                         "old_id>=$start", $fname, array( 'ORDER BY' => 'old_id', 'LIMIT' => $chunksize, 'FOR UPDATE' ) );
  21                 if( $dbw->numRows( $res ) == 0 ) {
  22                         break;
  23                 }
  24                 $last = $start;
  25                 while( $row = $dbw->fetchObject( $res ) ) {
  26                         # print "  {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
  27                         compressPage( $row );
  28                         $last = $row->old_id;
  29                 }
  30                 $dbw->freeResult( $res );
  31                 $start = $last + 1; # Deletion may leave long empty stretches
  32                 print "$start...\n";
  33         } while( true );
  34 }
  35
  36 function compressPage( $row ) {
  37         $fname = 'compressPage';
  38         if( false !== strpos( $row->old_flags, "gzip" ) ) {
  39                 print "Already compressed row {$row->old_id}?\n";
  40                 return false;
  41         }
  42         $dbw =& wfGetDB( DB_MASTER );
  43         $flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
  44         $compress = gzdeflate( $row->old_text );
  45         $dbw->update( 'old',
  46                 array( /* SET */
  47                         'old_flags' => $flags,
  48                         'old_text' => $compress
  49                 ), array( /* WHERE */
  50                         'old_id' => $row->old_id
  51                 ), $fname, 'LIMIT 1'
  52         );
  53         return true;
  54 }
  55
  56 define( 'LS_INDIVIDUAL', 0 );
  57 define( 'LS_CHUNKED', 1 );
  58
  59 function compressWithConcat( $startId, $maxChunkSize, $maxChunkFactor, $factorThreshold, $beginDate, $endDate )
  60 {
  61         $fname = 'compressWithConcat';
  62         $loadStyle = LS_CHUNKED;
  63
  64         $dbw =& wfGetDB( DB_MASTER );
  65
  66         # First get a list of all pages
  67         $pageRes = $dbw->select( 'page', 'page_id', false, $fname );
  68
  69         # For each of those, get a list of revisions which fit the criteria
  70         $conds = array();
  71         if ( $beginDate ) {
  72                 $conds[] = "rev_timestamp>'" . $beginDate . "'";
  73         }
  74         if ( $endDate )  {
  75                 $conds[] = "rev_timestamp<'" . $endDate . "'";
  76         }
  77         if ( $startId ) {
  78                 $conds[] = 'rev_id>=' . $startId;
  79         }
  80         if ( $loadStyle == LS_CHUNKED ) {
  81                 $tables = array( 'revision', 'text' );
  82                 $fields = array( 'rev_id', 'old_flags', 'old_text' );
  83                 $conds[] = 'rev_id=old_id';
  84                 $revLoadOptions = 'FOR UPDATE';
  85         } else {
  86                 $tables = array( 'revision' );
  87                 $fields = array( 'rev_id' );
  88                 $revLoadOptions = array();
  89         }
  90
  91         while ( $pageRow = $dbw->fetchObject( $pageRes ) ) {
  92                 # Display progress
  93                 $titleObj = Title::makeTitle( $pageRow->cur_namespace, $pageRow->cur_title );
  94                 print $titleObj->getPrefixedDBkey() . " ";
  95
  96                 # Load revisions
  97                 $revRes = $dbw->select( 'old', $fields,
  98                         array( 'rev_page' => $pageRow->page_id ) + $conds,
  99                         $fname,
 100                         $revLoadOptions
 101                 );
 102                 $revs = array();
 103                 while ( $revRow = $dbw->fetchObject( $revRes ) ) {
 104                         $revs[] = $revRow;
 105                 }
 106
 107                 if ( count( $revs ) < 2) {
 108                         # No revisions matching, no further processing
 109                         print "\n";
 110                         continue;
 111                 }
 112
 113                 # For each chunk
 114                 $i = 0;
 115                 while ( $i < count( $revs ) ) {
 116                         if ( $i < count( $revs ) - $maxChunkSize ) {
 117                                 $thisChunkSize = $maxChunkSize;
 118                         } else {
 119                                 $thisChunkSize = count( $revs ) - $i;
 120                         }
 121
 122                         $chunk = new ConcatenatedGzipHistoryBlob();
 123                         $stubs = array();
 124                         $dbw->begin();
 125                         $usedChunk = false;
 126                         $primaryOldid = $revs[$i]->rev_id;
 127
 128                         # Get the text of each revision and add it to the object
 129                         for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
 130                                 $oldid = $revs[$i + $j]->rev_id;
 131
 132                                 # Get text
 133                                 if ( $loadStyle == LS_INDIVIDUAL ) {
 134                                         $textRow = $dbw->selectRow( 'text',
 135                                                 array( 'old_flags', 'old_text' ),
 136                                                 array( 'old_id' => $oldid ),
 137                                                 $fname,
 138                                                 'FOR UPDATE'
 139                                         );
 140                                         $text = Revision::getRevisionText( $textRow );
 141                                 } else {
 142                                         $text = Revision::getRevisionText( $revs[$i + $j] );
 143                                 }
 144
 145                                 if ( $text === false ) {
 146                                         print "\nError, unable to get text in old_id $oldid\n";
 147                                         #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
 148                                 }
 149
 150                                 if ( $j == 0 ) {
 151                                         $chunk->setText( $text );
 152                                         print '.';
 153                                 } else {
 154                                         # Don't make a stub if it's going to be longer than the article
 155                                         # Stubs are typically about 100 bytes
 156                                         if ( strlen( $text ) < 120 ) {
 157                                                 $stub = false;
 158                                                 print 'x';
 159                                         } else {
 160                                                 $stub = $chunk->addItem( $text );
 161                                                 $stub->setLocation( $primaryOldid );
 162                                                 $hash = $stub->getHash();
 163                                                 $stub = serialize( $stub );
 164                                                 print '.';
 165                                                 $usedChunk = true;
 166                                         }
 167                                         $stubs[$j] = $stub;
 168                                 }
 169                         }
 170                         $thisChunkSize = $j;
 171
 172                         # If we couldn't actually use any stubs because the pages were too small, do nothing
 173                         if ( $usedChunk ) {
 174                                 # Store the main object
 175                                 $dbw->update( 'text',
 176                                         array( /* SET */
 177                                                 'old_text' => serialize( $chunk ),
 178                                                 'old_flags' => 'object',
 179                                         ), array( /* WHERE */
 180                                                 'old_id' => $primaryOldid
 181                                         )
 182                                 );
 183
 184                                 # Store the stub objects
 185                                 for ( $j = 1; $j < $thisChunkSize; $j++ ) {
 186                                         # Skip if not compressing
 187                                         if ( $stubs[$j] !== false ) {
 188                                                 $dbw->update( 'text',
 189                                                         array( /* SET */
 190                                                                 'old_text' => $stubs[$j],
 191                                                                 'old_flags' => 'object',
 192                                                         ), array( /* WHERE */
 193                                                                 'old_id' => $revs[$i + $j]->old_id
 194                                                         )
 195                                                 );
 196                                         }
 197                                 }
 198                         }
 199                         # Done, next
 200                         print "/";
 201                         $dbw->commit();
 202                         $i += $thisChunkSize;
 203                 }
 204                 print "\n";
 205         }
 206         return true;
 207 }
 208
 209 ?>